diff --git a/docs/knowledge-base.md b/docs/knowledge-base.md
index 7a6b9c36a..837652f24 100644
--- a/docs/knowledge-base.md
+++ b/docs/knowledge-base.md
@@ -14,7 +14,7 @@ separate work, not an implemented capability of this knowledge base.
 ```text
 src/inferrsi/data/skill_library/
   README.md                         Library navigation
-  skill-index.json                   Stable IDs → categorized paths/backend (76 skills)
+  skill-index.json                   Stable IDs → categorized paths/backend (77 skills)
   hardware/{ascend,cuda,host}/<id>/   Hardware/platform-specific complete skills
   frameworks/<framework>/<id>/      Framework-owned complete skills
   repositories/<repo>/              Engine entry, sub-skills and focused complete skills
@@ -171,18 +171,21 @@ Complete procedures, sidecars, references and cases are packaged, with 14 Ascend
 repositories, methods or workflows. The 21 engine and research-router skills
 use the same ownership layout under `repositories/<engine>/` and
 `workflows/inference-engine-research/`, including their sub-skills and references.
-The library contains 120 `SKILL.md` files, all in English.
+The library contains 124 `SKILL.md` files, all in English. The registered
+[Lithos Metal candidate](../src/inferrsi/data/skill_library/repositories/lithos-metal/SKILL.md)
+adds three owner-local subskills for compiler/fusion, generation state and
+serving/cache contracts; it adds no executable repository target.
 The collection overview, import notes and source manifest belong to the research
 workflow's references; Strata design-space analysis belongs to Strata. The library
 root contains only navigation and the registry.
 
 The [generated catalog](../src/inferrsi/data/repo_packs/inferrsi/knowledge/README.md)
-projects the 76 registered skills into the contribution catalog. Its records
+projects the 77 registered skills into the contribution catalog. Its records
 use `inferrsi.<backend>.skills.<name_with_underscores>` IDs, with `ascend` and
 `cuda` for scopes that depend on those backends and `common` for the remaining skills.
 Each `skill_documents.original` points to the full packaged procedure. The
 `skills` CLI retains the original hyphenated skill IDs and defaults to this
-76-ID registry; engine procedures remain available through library navigation
+77-ID registry; engine procedures remain available through library navigation
 and catalog document pointers.
 
 The registered [h3.c runtime audit](../src/inferrsi/data/skill_library/repositories/h3-c/audit-h3-runtime-residency/SKILL.md)
diff --git a/docs/reference-matrix.md b/docs/reference-matrix.md
index 7afb156a2..784cecbd9 100644
--- a/docs/reference-matrix.md
+++ b/docs/reference-matrix.md
@@ -47,6 +47,7 @@ Reading rules:
 | sgl-project/sglang @ `16176278d0180f1a1a578e53e6eef432577876c2` (Apache-2.0; conv adaptation notices credit vLLM and Tri Dao) | Bounded source inspection, 2026-10-29; Qwen3.5/Qwen3-Next caller, GDN phase/architecture policy, separate conv wrappers, state shapes/row mapping and selected upstream test intent; [readset and file identities](../src/inferrsi/data/skill_library/methods/kernel-writing-guide/references/sglang-gdn-causal-conv-contract.md#source-ledger-and-attribution) | mined (narrow) — per-runner prefill/decode, CUDA/Triton conv stride/alias branches, state write-back and conditional one-shot prefill transfer → [kernel-writing-guide](../src/inferrsi/data/skill_library/methods/kernel-writing-guide/SKILL.md), [four source cases](../src/inferrsi/data/skill_library/methods/kernel-writing-guide/cases/sglang-gdn-source-contract.md); [HF-worker/engine cross-map](../src/inferrsi/data/skill_library/methods/kernel-writing-guide/references/sglang-gdn-causal-conv-contract.md#map-the-inspected-hf-workers-to-vllm-and-sglang) adds separate [PPLX caller context](../src/inferrsi/data/skill_library/methods/kernel-writing-guide/references/pplx-gdn-caller-context.json) (worker license UNKNOWN); candidate/not_run, agent efficacy not_evaluated | out-of-scope — device state is not host RAM | out-of-scope | out-of-scope — row metadata is not network evidence | out-of-scope — engine phase classification is not OS scheduling |
 | NVlabs/kda @ `acdd79343c5e9d0c7cb96289f26fd3f1051ce785` → mit-han-lab/KernelWiki gitlink @ `76d27b56f804e7e7295d4c570e1e5d7eef4b0a75` | Direct bounded distillation, 2026-10-23; cutoff 2026-04-27. All 48 synthesis headers and 2,217 source headers screened; 171 SM90 source headers retained, selected bodies and independently matched DeepGEMM/CUTLASS code plus versioned NVIDIA sections inspected. [Readset/coverage/license qualifications](../src/inferrsi/data/skill_library/hardware/cuda/reason-hopper-kernels/references/sources.md); remaining PR bodies/algorithms unmined; KW pin has no root LICENSE, KDA MIT/sanitization assertions qualified | mined (conditional) — Hopper TMA/mbarrier/WGMMA ownership, resource/layout/pipeline/software scheduling and selected FP8 scaling → [reason-hopper-kernels](../src/inferrsi/data/skill_library/hardware/cuda/reason-hopper-kernels/SKILL.md); SM100 recipes and unsafe snippet generalizations excluded; native/precision/performance not_run | out-of-scope — device registers/SMEM/HBM are not host RAM | out-of-scope — no storage implementation extracted | out-of-scope — CTA/cluster synchronization is not network transport | out-of-scope — tile scheduling is not OS scheduling |
 | antirez/h3.c @ `8974cc055ea9c02fcd14cc27dfda3e1027c05153` (MIT; retained ccv BSD-3-Clause notice) | Direct bounded source distillation, 2026-10-23; 34 selected file bytes matched to pinned Git blobs; [readset/architecture](../src/inferrsi/data/skill_library/repositories/h3-c/audit-h3-runtime-residency/references/source-notes.md). Loader, generation/cache, TE/vision/input-output VAEs, AdaLN, selected Metal/int8/SSD paths and native test assertions; full kernels/driver/media internals unread. Prior model-header observations reused, not rerun | mined — joint audio/video schedule and row-map dependencies, BF16 rounding/fusion versus numerical approximation, actual TensorOps dispatch → [audit-h3-runtime-residency](../src/inferrsi/data/skill_library/repositories/h3-c/audit-h3-runtime-residency/SKILL.md); native/quality/performance not_run | mined — shared/file-backed weights, Qwen prefetch ring, phase/cache/preview overlap and complete host output; tensor counters are not physical footprint or 64 GiB qualification | mined — exact raw BF16/F32 schema, direct/mapped reads and two-slot BF16 matrix streaming with completion/join boundaries; uncached-read request is not measured cache bypass or useful overlap | out-of-scope — local GPU/I/O ordering is not network transport | out-of-scope — no OS scheduling/control implementation extracted |
+| lithos-ai/lithos-metal @ `6653b7df5f2c4392ec7b038960f2796b1499923f` (Apache-2.0; notices retained) | Bounded compiler/Program/StepState/DSpark, backend identity and serving/pack/prefix/idle paths; selected contract/runtime assertions; [readset](../src/inferrsi/data/skill_library/repositories/lithos-metal/references/source-notes.md). Candidate; software/CPU/synthetic Metal checks do not qualify full-model numerics, DSpark or service performance | mined — chip-qualified task fusion, normalized controls, hazards/scratch and committed generation state → [lithos-metal](../src/inferrsi/data/skill_library/repositories/lithos-metal/SKILL.md); empirical efficacy not_run | mined — program/pipeline versus GPU allocation and exact host-prefix ownership; unified-memory residency/performance unqualified | mined — derived pack validity, exact-prefix disk identity/integrity, bounded persistence and quota; weight-byte authentication and crash durability not established | out-of-scope — HTTP adapters are not distributed transport | out-of-scope — idle lifecycle is not OS scheduling |
 | zuiho-kai/h3-metal @ `4ba47a3eabdbebf801ac6664b31f8ec168bc57a8` (inspected MIT; retained BSD-3-Clause notice; machine resolver unavailable) | Bounded source distillation, 2026-10-28; 21 identity-checked files with selected ranges; [scope/readset](../src/inferrsi/data/skill_library/repositories/h3-metal/audit-h3-metal-fasth3-semantics/references/source-notes.md). Fork-specific FastH3 params, tensor/schedule/RoPE and VSA dispatch, test/script limits; full engine/external FastVideo/model surface not mined | mined — requested versus effective controls, ambient conflicts, head-interleaved QKV/FC1 swap/BF16 schedule, real partial VSA tiles/double top-k and dense-plus-coarse fallback → [audit-h3-metal-fasth3-semantics](../src/inferrsi/data/skill_library/repositories/h3-metal/audit-h3-metal-fasth3-semantics/SKILL.md); native/numerical/quality/performance not_run | out-of-scope — tile arithmetic is not measured host footprint or memory-fit qualification | out-of-scope — FastH3 disables requested SSD parameter; no new storage optimization or payload validation | out-of-scope — local attention/dispatch is not network transport | out-of-scope — runtime selection is not OS scheduling |
 | Comfy-Org/ComfyUI @ `b1a247c9cf9dbfda4b2c0a0c1d50950667eeaa0c` (GNU GPL v3) | Source-reading archive, inspected 2026-10-21; bounded execution/cache/graph/model-management/static-dynamic-patcher/sampler paths and selected tests; [readset](../src/inferrsi/data/skill_library/repositories/comfy-ui/references/repo-provenance.md); not full-model/custom-node/provider-native coverage | mined — managed weight/patch identity, partial unload versus detach and dynamic-backend boundary → [comfy-ui](../src/inferrsi/data/skill_library/repositories/comfy-ui/SKILL.md); device behavior unverified | mined — output versus object identity, ancestor fingerprints, lazy consumer references and retention versus validity; host/pin ownership | out-of-scope — no storage implementation extracted | out-of-scope — no transport implementation extracted | out-of-scope — workflow scheduling is not OS scheduling |
 | leejet/stable-diffusion.cpp @ `b13a48219d5cdf408edcb93da6cd69b612ed03d3` (MIT) | Supplied archive, inspected 2026-10-22: 18 bounded paths plus separately identity-checked README (19 blobs); historical receipt preserved. 2026-10-23 extension: 12 selected C++ blobs plus retained MIT text identity-checked, with line-scoped catalog/non-ZIP safetensors read/staging/mapping and immediate consumer cleanup; [readsets](../src/inferrsi/data/skill_library/repositories/stable-diffusion-cpp/audit-sd-cpp-residency/references/source-notes.md). File identity is not full-file review; archive/submodules, model kernels and other-format implementations remain outside the bounded audit | mined — native weight/runtime capacity, protected segment/prefetch lifetime → [audit-sd-cpp-residency](../src/inferrsi/data/skill_library/repositories/stable-diffusion-cpp/audit-sd-cpp-residency/SKILL.md); native/backend behavior unverified | mined — parameter storage versus compute staging and retained runtime ownership; loader host read/conversion buffers versus mapped owners; host peaks unmeasured | mined — bounded catalog stamps versus content identity, range admission versus actual reads, mapping/shared-owner and newly created block cleanup → [loader/storage boundaries](../src/inferrsi/data/skill_library/repositories/stable-diffusion-cpp/audit-sd-cpp-residency/references/loader-storage-boundaries.md); full other-format parsing, native failure/SSD/performance qualification remain open | out-of-scope — device copies are not network transport | out-of-scope — runner scheduling is not OS scheduling |
diff --git a/docs/searchspace-matrix.md b/docs/searchspace-matrix.md
index 21ff570d9..a483aed3e 100644
--- a/docs/searchspace-matrix.md
+++ b/docs/searchspace-matrix.md
@@ -27,6 +27,7 @@ The primary compute and HBM resource. Most current work lives here.
 | Lifecycle step | What it covers | Candidate Skills | Cases / Experience | Evidence | Maturity |
 |---|---|---|---|---|---|
 | **Diagnose / Configure / Verify** | CUDA eager dense-FFN Attention/FFN rank namespaces and per-stage/per-peer hidden-state ownership | [audit-afd-p2p-exchanges](../src/inferrsi/data/skill_library/repositories/afd-plugin/audit-afd-p2p-exchanges/SKILL.md) | A=3,F=2 control sender differs from data peer; matching total tokens does not preserve peer rows | afd-plugin @ df8d6a0c403c3ca1b4ed9cf1687e7204908bebec; [source ledger](../src/inferrsi/data/skill_library/repositories/afd-plugin/audit-afd-p2p-exchanges/references/source-notes.md), [proposed cases](reviews/issue91-afd-p2p.md) | 🟢 (source-derived candidate; native/model/efficacy not_run; no Network qualification) |
+| **Diagnose / Configure / Optimize / Verify** | Chip-qualified task fusion, normalized-versus-fused controls, ICB hazards/scratch and GPU-resident accepted-prefix state | [lithos-metal](../src/inferrsi/data/skill_library/repositories/lithos-metal/SKILL.md), [compile-fusion](../src/inferrsi/data/skill_library/repositories/lithos-metal/sub-skills/compile-fusion/SKILL.md), [generation-state](../src/inferrsi/data/skill_library/repositories/lithos-metal/sub-skills/generation-state/SKILL.md) | Layer mixer and MLP regions differ; CPU/synthetic ICB checks cannot establish full-model/DSpark numerics or E2E rates | Lithos Metal @ `6653b7df5f2c4392ec7b038960f2796b1499923f`; bounded source readset | Source-derived candidate; efficacy not_run |
 | **Observe / Diagnose / Verify** | MiniMax-H3 joint audio/video AdaLN schedule/row-map validity, Metal BF16 rounding and actual int8 dispatch versus approximate controls | [audit-h3-runtime-residency](../src/inferrsi/data/skill_library/repositories/h3-c/audit-h3-runtime-residency/SKILL.md) | Fifty layers/reuse 1 does not disable M5 int8 or BF16 head; test-source parity is not full media quality | h3.c @ 8974cc055ea9c02fcd14cc27dfda3e1027c05153; [precision contracts](../src/inferrsi/data/skill_library/repositories/h3-c/audit-h3-runtime-residency/references/precision-and-streaming.md) | 🟢 (source-derived candidate; native/quality/performance not_run) |
 | **Observe / Diagnose / Verify** | FastH3 effective parameters/ambient override, semantic checkpoint transformations, partial/prefix VSA membership and sparse eligibility/fallback | [audit-h3-metal-fasth3-semantics](../src/inferrsi/data/skill_library/repositories/h3-metal/audit-h3-metal-fasth3-semantics/SKILL.md) | Shape agreement is not conversion; sparse-disabled remains dense fine+coarse; parameter normalization does not erase ambient token enable | h3-metal @ 4ba47a3eabdbebf801ac6664b31f8ec168bc57a8; [semantic counterchecks](../src/inferrsi/data/skill_library/repositories/h3-metal/audit-h3-metal-fasth3-semantics/references/semantic-contracts.md) | 🟢 (source-derived candidate; native/sparse/schedule/media/performance not_run) |
 | **Diagnose / Configure / Verify** | Workflow model-weight residency, partial unload versus wrapper removal, patch identity and static/dynamic backend dispatch | [comfy-ui](../src/inferrsi/data/skill_library/repositories/comfy-ui/SKILL.md) | CPU dynamic construction reroutes to the static patcher; wrapper/free-byte bookkeeping is not physical allocator release | ComfyUI @ b1a247c9cf9dbfda4b2c0a0c1d50950667eeaa0c; [model ownership](../src/inferrsi/data/skill_library/repositories/comfy-ui/references/model-residency.md) | 🟢 (source-derived audit; device/native allocator unverified) |
@@ -85,6 +86,7 @@ Host-side memory: CPU DRAM, pinned memory, page tables, PyTorch tensor lifecycle
 
 | Lifecycle step | What it covers | Candidate Skills | Cases / Experience | Evidence | Maturity |
 |---|---|---|---|---|---|
+| **Diagnose / Configure / Verify** | Program/pipeline cache, GPU allocation adoption, exact host-prefix snapshots and pending-request idle release | [lithos-metal serving-cache](../src/inferrsi/data/skill_library/repositories/lithos-metal/sub-skills/serving-cache/SKILL.md) | Allocation reuse is separate from sequence validity; host snapshots and OS file cache have distinct owners | Lithos Metal pinned serving/Session/PrefixCache/lifecycle paths | Source-derived candidate; full-model residency and continuation unverified |
 | **Observe / Diagnose / Verify** | MiniMax-H3 TE/vision/input-output VAE phase lifetime, prepared DiT/decoder retention, shared/mapped weight and host media/output residency | [audit-h3-runtime-residency](../src/inferrsi/data/skill_library/repositories/h3-c/audit-h3-runtime-residency/SKILL.md) | Preview/cache changes overlap; output tiling retains full RGB; checkpoint bytes and tracked tensors do not establish total RAM fit | h3.c @ 8974cc055ea9c02fcd14cc27dfda3e1027c05153; [phase ledger](../src/inferrsi/data/skill_library/repositories/h3-c/audit-h3-runtime-residency/references/phase-residency.md) | 🟢 (source-derived candidate; physical/compressed/swap peaks unmeasured) |
 | **Diagnose / Configure / Verify** | Workflow-output signatures and object identity, ancestor/fingerprint validity versus retention, lazy consumer references and host/pinned state | [comfy-ui](../src/inferrsi/data/skill_library/repositories/comfy-ui/SKILL.md) | A fixed JSON filename is not an external-state fingerprint; NOT_IDEMPOTENT node identity does not force every-run invalidation | ComfyUI @ b1a247c9cf9dbfda4b2c0a0c1d50950667eeaa0c; [cache ledger](../src/inferrsi/data/skill_library/repositories/comfy-ui/references/graph-cache-invalidation.md) | 🟢 (source-derived procedure; executor cache suite not run) |
 | **Observe / Diagnose / Verify** | Native diffusion parameter storage, compute staging and retained runtime buffers across conditioning/denoise/VAE phases; loader read/conversion buffers and shared mapping owners | [audit-sd-cpp-residency](../src/inferrsi/data/skill_library/repositories/stable-diffusion-cpp/audit-sd-cpp-residency/SKILL.md) | Device fit does not establish sufficient host RAM; initial host preparation may precede async transfer; releasing one owner is not final unmap; track host/device peaks separately | stable-diffusion.cpp @ b13a48219d5cdf408edcb93da6cd69b612ed03d3; [validation](../src/inferrsi/data/skill_library/repositories/stable-diffusion-cpp/audit-sd-cpp-residency/references/validation-boundaries.md), [loader/storage](../src/inferrsi/data/skill_library/repositories/stable-diffusion-cpp/audit-sd-cpp-residency/references/loader-storage-boundaries.md) | 🟢 (source-derived ownership audit; host peaks unmeasured) |
@@ -115,6 +117,7 @@ Storage I/O: model weight loading, checkpoint save/restore, profiling artifact I
 | Lifecycle step | What it covers | Candidate Skills | Cases / Experience | Evidence | Maturity |
 |---|---|---|---|---|---|
 | **Diagnose / Verify** | Native diffusion catalog/source identity → non-ZIP safetensors byte reads → host staging or retained mapping → immediate ModelManager completion/cleanup | [audit-sd-cpp-residency](../src/inferrsi/data/skill_library/repositories/stable-diffusion-cpp/audit-sd-cpp-residency/SKILL.md) | Stamps/internal revisions are not content hashes; valid metadata or a destination callback is not read completion; failure cleanup covers this load's newly created blocks | stable-diffusion.cpp @ b13a48219d5cdf408edcb93da6cd69b612ed03d3; [loader/storage boundaries](../src/inferrsi/data/skill_library/repositories/stable-diffusion-cpp/audit-sd-cpp-residency/references/loader-storage-boundaries.md) | 🟢 (bounded source-derived candidate; other formats, native I/O and performance unqualified) |
+| **Diagnose / Configure / Verify** | Prepared-pack structural identity and exact-prefix disk blobs, hashes, quota and bounded flush | [lithos-metal serving-cache](../src/inferrsi/data/skill_library/repositories/lithos-metal/sub-skills/serving-cache/SKILL.md) | Weight payload size/mtime is not byte authentication; queued writes are not persistence; corrupt/stale entries prefill | Lithos Metal pinned pack/PrefixStore paths | Source-derived candidate; durability/performance unverified |
 | **Diagnose / Configure / Verify** | Raw tensor-schema compatibility, file-backed DiT versus two-slot BF16 SSD streaming, Qwen read-worker/ring-depth distinction and submit/join completion | [audit-h3-runtime-residency](../src/inferrsi/data/skill_library/repositories/h3-c/audit-h3-runtime-residency/SKILL.md) | SSD mode disables runtime int8; compare matching BF16 paths. F_NOCACHE request and zero join wait do not prove bypass or free I/O | h3.c @ 8974cc055ea9c02fcd14cc27dfda3e1027c05153; [I/O tradeoffs](../src/inferrsi/data/skill_library/repositories/h3-c/audit-h3-runtime-residency/references/precision-and-streaming.md#three-different-ioresidency-paths) | 🟢 (source-derived candidate; native I/O/overlap/performance not_run) |
 | **Observe / Profile** | Workload-matched block-device measurement (fio methodology: signature recording, requested-vs-achieved depth, cache-state ledger, run-shape classification), per-device attribution (iostat), per-process attribution (pidstat) | `storage-io-measurement`, `weight-loading-optimization` (§7 phase timers) | Bounded observation templates; unit/latency-provenance schema; logical-vs-physical I/O distinction | fio @ acb7a302 + sysstat @ efd0c445 (pack-mediated, wave 6); fastsafetensors perf.py @ 219e256; mindie-agent docs @ 6a81b41 | 🟢 (methodology only — running benchmarks or manipulating caches requires separate authorization) |
 | **Diagnose** | Slow startup (graph capture + weight loading dominating): distinguish compile warmup from I/O | `ascend-graph-mode` (mentions slow startup), `ascend-model-adaptation` (dummy→real sequence) | ACLGraph startup cost is compile + capture, not I/O bound; FP8 weight loading pitfalls | MindIE-SD; vllm-ascend | 🟡 (incidental coverage only) |
diff --git a/docs/skill-coverage.md b/docs/skill-coverage.md
index 8bc366b59..b565fd927 100644
--- a/docs/skill-coverage.md
+++ b/docs/skill-coverage.md
@@ -4,7 +4,7 @@ Generated from the registry, the skill sidecars and the two maintained
 matrices. **Do not edit by hand**; regenerate after any registry, skill,
 matrix or frozen-plan change.
 
-- Input digest: `sha256:fd54d787da07b208e2d4b468a099f182359cbed8713bea9cdff3f9544f948a01`
+- Input digest: `sha256:7da50a38ecca8952f735441cb89758f4bd5893f8c97e023f9ef489b194e8a419`
 - Regenerate: `python tools/build_skill_coverage.py`
 - Verify without writing: `python tools/build_skill_coverage.py --check`
 
@@ -31,17 +31,17 @@ matrix: every field is copied from an owner that already maintains it.
 
 | Measure | Value |
 |---|---|
-| Registered skill ids | 76 |
-| By category | frameworks 3, hardware 22, methods 20, repositories 25, workflows 6 |
-| By backend | ascend 14, common 57, cuda 5 |
-| Recorded on a numbered SearchSpace resource layer | 48 |
+| Registered skill ids | 77 |
+| By category | frameworks 3, hardware 22, methods 20, repositories 26, workflows 6 |
+| By backend | ascend 14, common 58, cuda 5 |
+| Recorded on a numbered SearchSpace resource layer | 49 |
 | Recorded only in a supplementary guidance table (†) | 13 |
 | Recorded on two or more resource layers | 24 |
-| Recorded in the Reference Matrix | 68 |
-| With a `metadata.json` sidecar | 71 |
+| Recorded in the Reference Matrix | 69 |
+| With a `metadata.json` sidecar | 72 |
 | Selected by a frozen evaluation plan | 2 |
-| `SKILL.md` documents in the library | 120 |
-| Unregistered documents (excluded from every count above) | 44 |
+| `SKILL.md` documents in the library | 124 |
+| Unregistered documents (excluded from every count above) | 47 |
 
 A complete inventory is not a coverage claim. Every registered row below
 has an entry; that says the listing is exhaustive, not that inference
@@ -82,7 +82,7 @@ optimization is fully covered or that any procedure has measured benefit.
 | [diffusers-invocation-state](../src/inferrsi/data/skill_library/frameworks/diffusers/diffusers-invocation-state/SKILL.md) | common | Trace Diffusers pipeline-wrapper and scheduler invocation state when repeated calls, from_pipe reuse or interruption ... | 2 RAM/Host † guidance: Diffusers invocation guidance (source-derived) | Diffusers invocation source extraction (huggingface/diffusers @ `aea7b70be0f594bcdc15a6f2677ee2ca85e49e8e` ...) | candidate · not_run · 1 readset(s) | no-eval |
 | [lazy-evaluation-stream-lifetimes](../src/inferrsi/data/skill_library/frameworks/mlx/lazy-evaluation-stream-lifetimes/SKILL.md) | common | Diagnose MLX lazy evaluation, cross-stream ordering and array-buffer lifetimes when timing, memory use or asynchronous ... | 1 Accelerator † guidance: MLX framework guidance (source-derived) | MLX framework source extraction (ml-explore/mlx @ `bff10220c742907f0c68e55ff174aaea56f74e3d` (MIT)) | candidate · not_run · 1 readset(s) | no-eval |
 
-## Registered: Repositories (25)
+## Registered: Repositories (26)
 
 | Skill | Backend | Declared task | SearchSpace layer / stage | Reference Matrix | State | Gaps |
 |---|---|---|---|---|---|---|
@@ -104,6 +104,7 @@ optimization is fully covered or that any procedure has measured benefit.
 | [husky](../src/inferrsi/data/skill_library/repositories/husky/SKILL.md) | common | Review Husky's official-document claims as research hypotheses, separating author reports, implementation evidence ... | — | Husky document-only research adaptation (Official https://husky.underdog.ai/; displayed2026-09-20 ...) | candidate · not_run · 2 readset(s) | no-SearchSpace-layer, no-eval |
 | [kv-tier-management](../src/inferrsi/data/skill_library/repositories/lmcache/kv-tier-management/SKILL.md) | common | Manage runtime KV cache tiers (GPU → CPU RAM → local disk) for LLM serving: decide when KV is stored, retained ... | 2 RAM/Host · Configure / Optimize<br>3 SSD · Optimize | 1 · Systems sources (LMCache/LMCache @ c3b875a) | candidate · not_run · 0 readset(s) | no-eval |
 | [lily](../src/inferrsi/data/skill_library/repositories/lily/SKILL.md) | common | Investigate Lily's exact Qwen3.6-35B-A3B MLX-affine-Q4 contract, Metal state and kernels, strict greedy text API ... | — | Lily portable AREX extraction (perplexityai/pplx-garden `lily/` @ `1ed972ed3f0bd5616c997c9507c25616c6 ...) | candidate · not_run · 2 readset(s) | no-SearchSpace-layer, no-eval |
+| [lithos-metal](../src/inferrsi/data/skill_library/repositories/lithos-metal/SKILL.md) | common | Inspect Lithos Metal compilation and fusion, GPU-resident generation and DSpark state, or local serving and ... | 1 Accelerator · Diagnose / Configure / Optimize / Verify | 1 · Systems sources (lithos-ai/lithos-metal @ `6653b7df5f2c4392ec7b038960f2796b1499923f` ...) | candidate · not_run · 1 readset(s) | no-eval |
 | [local-dream](../src/inferrsi/data/skill_library/repositories/local-dream/SKILL.md) | common | Audit Local Dream Android desired-versus-serving process state, remote exposure, native backend and DiT packaging/ABI ... | 2 RAM/Host † guidance: Local Dream Android diffusion source contracts | Local Dream Android diffusion boundary source (xororz/local-dream @ `db162df4d9bb6efc6595250423d8df26241f4c27` ...) | candidate · not_run · 2 readset(s) | no-eval |
 | [mistral-rs](../src/inferrsi/data/skill_library/repositories/mistral-rs/SKILL.md) | common | Investigate mistral.rs Metal paged-attention decode layouts, prefill coordinates and workspace, actual-model prewarm ... | 1 Accelerator † guidance: mistral.rs Metal paged-attention source contracts<br>2 RAM/Host † guidance: mistral.rs Metal paged-attention source contracts | mistral.rs Metal paged-attention extraction (EricLBuehler/mistral.rs @ `9459891e7697238225885518119682ac8209a612` ...) | candidate · not_run · 2 readset(s) | no-eval |
 | [mlc-llm](../src/inferrsi/data/skill_library/repositories/mlc-llm/SKILL.md) | common | Investigate MLC LLM compiled Metal artifact/schema, Apple target and pinned TVM module/storage/stream contracts, and ... | 1 Accelerator † guidance: MLC LLM artifact, target and runtime source contracts<br>2 RAM/Host † guidance: MLC LLM artifact, target and runtime source contracts | MLC LLM compiled Metal and runtime contracts (mlc-ai/mlc-llm @ `8978ea9aec626d64096aaa92fbc89bda2e3ea802`; TVM ...) | candidate · not_run · 2 readset(s) | no-eval |
@@ -150,7 +151,7 @@ optimization is fully covered or that any procedure has measured benefit.
 
 ## Documents outside the registry
 
-44 `SKILL.md` documents are not registered, so they have
+47 `SKILL.md` documents are not registered, so they have
 no stable id and no row above. They are engine entrypoints, their
 sub-skills, and the research router. Counting them would inflate coverage:
 an entrypoint routes to procedures below it rather than adding one.
@@ -163,6 +164,9 @@ an entrypoint routes to procedures below it rather than adding one.
 | `repositories/lily/sub-skills/model-runtime` | sub-skill |
 | `repositories/lily/sub-skills/qualification` | sub-skill |
 | `repositories/lily/sub-skills/serving-cache` | sub-skill |
+| `repositories/lithos-metal/sub-skills/compile-fusion` | sub-skill |
+| `repositories/lithos-metal/sub-skills/generation-state` | sub-skill |
+| `repositories/lithos-metal/sub-skills/serving-cache` | sub-skill |
 | `repositories/mistral-rs/sub-skills/decode-contract` | sub-skill |
 | `repositories/mistral-rs/sub-skills/prefill-contract` | sub-skill |
 | `repositories/mistral-rs/sub-skills/qualification` | sub-skill |
@@ -204,7 +208,7 @@ an entrypoint routes to procedures below it rather than adding one.
 
 ## Integrity checks
 
-- Registered ids resolve to an existing `SKILL.md`: 76 of 76.
+- Registered ids resolve to an existing `SKILL.md`: 77 of 77.
 - Duplicate ids or paths: none; `skill_catalog` rejects them before this view is built.
 - Skill links in the two matrices that do not resolve: 0.
 - Registered skills without a `metadata.json` sidecar: 5.
@@ -216,7 +220,7 @@ an entrypoint routes to procedures below it rather than adding one.
 | `no-SearchSpace-layer` | 15: ascend-video-diffusion-optimization, audit-fixed-workload-claim, audit-recurrent-kv-finalization, cite-check-downstream-pr, competitor-scan, draft-acceptance-issue, dwarfstar, husky, inference-benchmark-design, lily, npu-skill-factory, repository-evolution-analysis, skill-evolution-playbook, splash, video-diffusion-optimization |
 | `no-Reference-Matrix-row` | 8: ascend-video-diffusion-optimization, audit-fixed-workload-claim, cite-check-downstream-pr, claim-contract, competitor-scan, draft-acceptance-issue, rsi-evidence, video-diffusion-optimization |
 | `no-sidecar` | 5: cite-check-downstream-pr, claim-contract, competitor-scan, draft-acceptance-issue, rsi-evidence |
-| `no-eval` | 74: as listed in the Gaps column above |
+| `no-eval` | 75: as listed in the Gaps column above |
 
 These are inventory gaps, not defects. A workflow skill may legitimately
 have no hardware resource layer; a candidate may legitimately be untested.
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/README.md b/src/inferrsi/data/repo_packs/inferrsi/knowledge/README.md
index 089906de8..9cc49512c 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/README.md
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/README.md
@@ -1,6 +1,6 @@
 # InferRSI skill discovery index
 
-Generated abstracts for all 76 registered skills in `data/skill_library/`.
+Generated abstracts for all 77 registered skills in `data/skill_library/`.
 Their complete procedures, references and cases ship in the same package.
 `skill-index.json` maps stable skill IDs to categorized directories. The
 reference and searchspace matrices
@@ -15,7 +15,7 @@ These are knowledge abstracts (`--kind knowledge`), with namespaced package
 IDs distinct from explicitly selected `checkout_ref` skills. Ascend/CUDA
 procedures use their corresponding backend; shared methods use `common`.
 
-Source snapshot: [2f72a0ccabd87eefdc76223c9b82f7091c379da1](https://github.com/InferMatrixLab/InferRSI/tree/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library).
+Source snapshot: [26743abcfae48cd6223bd757e74f193d97e3e5a7](https://github.com/InferMatrixLab/InferRSI/tree/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library).
 
 ```bash
 inferrsi knowledge search 'profiling' --repo-type inferrsi --limit 50
@@ -29,7 +29,7 @@ contain the current input paths and bytes (Git checkout filters are respected).
 
 ```bash
 python tools/build_skill_index.py --source-revision <committed-source-revision>
-python tools/build_skill_index.py --source-revision 2f72a0ccabd87eefdc76223c9b82f7091c379da1 --check
+python tools/build_skill_index.py --source-revision 26743abcfae48cd6223bd757e74f193d97e3e5a7 --check
 ```
 
 `--check` detects missing, stale and unexpected generated records without writing.
@@ -90,6 +90,7 @@ References and cases ship with their owning skill, not as duplicate knowledge ab
 | [kv-tier-management](common/skills/kv-tier-management.json) | common | candidate |
 | [lazy-evaluation-stream-lifetimes](common/skills/lazy-evaluation-stream-lifetimes.json) | common | candidate |
 | [lily](common/skills/lily.json) | common | candidate |
+| [lithos-metal](common/skills/lithos-metal.json) | common | candidate |
 | [local-dream](common/skills/local-dream.json) | common | candidate |
 | [mistral-rs](common/skills/mistral-rs.json) | common | candidate |
 | [mlc-llm](common/skills/mlc-llm.json) | common | candidate |
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/ascend-aclgraph-capture-replay.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/ascend-aclgraph-capture-replay.json
index 20859c339..ad8322703 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/ascend-aclgraph-capture-replay.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/ascend-aclgraph-capture-replay.json
@@ -21,8 +21,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/hardware/ascend/ascend-aclgraph-capture-replay/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/hardware/ascend/ascend-aclgraph-capture-replay/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/hardware/ascend/ascend-aclgraph-capture-replay/SKILL.md",
         "src/inferrsi/data/skill_library/hardware/ascend/ascend-aclgraph-capture-replay/metadata.json",
@@ -41,7 +41,7 @@
     "src/inferrsi/data/skill_library/hardware/ascend/ascend-aclgraph-capture-replay/SKILL.md": "d494fc7e539cbb7388189bd7b3f04bb5c4983fdd7a82484348aad1cee7f741f2",
     "src/inferrsi/data/skill_library/hardware/ascend/ascend-aclgraph-capture-replay/metadata.json": "71c68431663ed2336cccc985a354adab6b13b4cfd6123a18a5d8d53c28268509",
     "src/inferrsi/data/skill_library/hardware/ascend/ascend-aclgraph-capture-replay/references/source-notes.md": "6ffbfb68cc7e04923b4fbca2a73a7555b7c3b45a6f40819bb5f3213ef05cf88d",
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190"
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
   },
   "skill_metadata": {
     "name": "ascend-aclgraph-capture-replay",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/ascend-device-health.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/ascend-device-health.json
index 047a45079..0a8b2f197 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/ascend-device-health.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/ascend-device-health.json
@@ -23,8 +23,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/hardware/ascend/ascend-device-health/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/hardware/ascend/ascend-device-health/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/hardware/ascend/ascend-device-health/SKILL.md",
         "src/inferrsi/data/skill_library/hardware/ascend/ascend-device-health/metadata.json",
@@ -41,7 +41,7 @@
   "source_files": {
     "src/inferrsi/data/skill_library/hardware/ascend/ascend-device-health/SKILL.md": "03829039860ece35519f2e4c339a48b5014ec25e07a2c150a3efb6a3a4e35b4a",
     "src/inferrsi/data/skill_library/hardware/ascend/ascend-device-health/metadata.json": "80d80c15514919826fa72d6f6cf26b76db8accd85d21187fc239232ca454e7bf",
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190"
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
   },
   "skill_metadata": {
     "name": "ascend-device-health",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/ascend-distributed-parallel.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/ascend-distributed-parallel.json
index d0b7fe531..d623c3cc1 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/ascend-distributed-parallel.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/ascend-distributed-parallel.json
@@ -24,8 +24,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/hardware/ascend/ascend-distributed-parallel/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/hardware/ascend/ascend-distributed-parallel/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/hardware/ascend/ascend-distributed-parallel/SKILL.md",
         "src/inferrsi/data/skill_library/hardware/ascend/ascend-distributed-parallel/metadata.json",
@@ -44,7 +44,7 @@
     "src/inferrsi/data/skill_library/hardware/ascend/ascend-distributed-parallel/SKILL.md": "4ce9dcadd1ae3c7a8cc633336760d0a3c382f354f1b7028e3a10d28fa0d0c9fa",
     "src/inferrsi/data/skill_library/hardware/ascend/ascend-distributed-parallel/metadata.json": "b6cb25879b1bca5f5af08b48776f2e664cf5dcd11a0efb1c4f56879e2afa9006",
     "src/inferrsi/data/skill_library/hardware/ascend/ascend-distributed-parallel/references/awesome-ascend-communication-tests.md": "0cb27a3d24409171bb414d7ebf0c49ad62be5f46482f29624695496ba9eb3e63",
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190"
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
   },
   "skill_metadata": {
     "name": "ascend-distributed-parallel",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/ascend-graph-mode.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/ascend-graph-mode.json
index 2c0b27c48..c33277c3d 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/ascend-graph-mode.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/ascend-graph-mode.json
@@ -22,8 +22,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/hardware/ascend/ascend-graph-mode/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/hardware/ascend/ascend-graph-mode/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/hardware/ascend/ascend-graph-mode/SKILL.md",
         "src/inferrsi/data/skill_library/hardware/ascend/ascend-graph-mode/metadata.json",
@@ -40,7 +40,7 @@
   "source_files": {
     "src/inferrsi/data/skill_library/hardware/ascend/ascend-graph-mode/SKILL.md": "9203cc5a14a1d65df5fbf275734b93be023e06527459b2f79864f28a338cb44a",
     "src/inferrsi/data/skill_library/hardware/ascend/ascend-graph-mode/metadata.json": "72b41e7845915d34b09219624e82976386f3125816293540c2c38143fe6b477f",
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190"
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
   },
   "skill_metadata": {
     "name": "ascend-graph-mode",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/ascend-model-adaptation.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/ascend-model-adaptation.json
index 0a9b08590..9ef69b816 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/ascend-model-adaptation.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/ascend-model-adaptation.json
@@ -22,8 +22,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/hardware/ascend/ascend-model-adaptation/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/hardware/ascend/ascend-model-adaptation/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/hardware/ascend/ascend-model-adaptation/SKILL.md",
         "src/inferrsi/data/skill_library/hardware/ascend/ascend-model-adaptation/metadata.json",
@@ -40,7 +40,7 @@
   "source_files": {
     "src/inferrsi/data/skill_library/hardware/ascend/ascend-model-adaptation/SKILL.md": "8fdf46fb2743b67a66bc069d22a45305a222debfb606777bc451b4b8f31d9cc6",
     "src/inferrsi/data/skill_library/hardware/ascend/ascend-model-adaptation/metadata.json": "be681a3206e5e183b4c9670c9d7784bcffce18c891449e5b15f94ca5f1324b9a",
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190"
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
   },
   "skill_metadata": {
     "name": "ascend-model-adaptation",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/ascend-operator-profiling.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/ascend-operator-profiling.json
index 86b33f321..39a549d03 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/ascend-operator-profiling.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/ascend-operator-profiling.json
@@ -23,8 +23,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/hardware/ascend/ascend-operator-profiling/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/hardware/ascend/ascend-operator-profiling/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/hardware/ascend/ascend-operator-profiling/SKILL.md",
         "src/inferrsi/data/skill_library/hardware/ascend/ascend-operator-profiling/metadata.json",
@@ -41,7 +41,7 @@
   "source_files": {
     "src/inferrsi/data/skill_library/hardware/ascend/ascend-operator-profiling/SKILL.md": "401981cf7828edcab0c9ea882f7ecd83badfa3c2a95006e912cf99d67b3d6a12",
     "src/inferrsi/data/skill_library/hardware/ascend/ascend-operator-profiling/metadata.json": "efd37bb23acb500dad4ec8ebe7ddfe5c3f196374052825f923ac85da8445f06b",
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190"
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
   },
   "skill_metadata": {
     "name": "ascend-operator-profiling",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/ascend-operator-tuning.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/ascend-operator-tuning.json
index d2d276c85..66c606e2a 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/ascend-operator-tuning.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/ascend-operator-tuning.json
@@ -24,8 +24,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/hardware/ascend/ascend-operator-tuning/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/hardware/ascend/ascend-operator-tuning/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/hardware/ascend/ascend-operator-tuning/SKILL.md",
         "src/inferrsi/data/skill_library/hardware/ascend/ascend-operator-tuning/metadata.json",
@@ -42,7 +42,7 @@
   "source_files": {
     "src/inferrsi/data/skill_library/hardware/ascend/ascend-operator-tuning/SKILL.md": "38b151bc6b4e9734c6fa41efb7a30343559400c0c75bbdd26bd8a968b8ea4592",
     "src/inferrsi/data/skill_library/hardware/ascend/ascend-operator-tuning/metadata.json": "04d2002fc8a0a7becf86b0b73daaf2da8716ab466d7f988da43d41a10f26680e",
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190"
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
   },
   "skill_metadata": {
     "name": "ascend-operator-tuning",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/ascend-optimization-casebook.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/ascend-optimization-casebook.json
index b3c26e016..4bf8bec5d 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/ascend-optimization-casebook.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/ascend-optimization-casebook.json
@@ -22,8 +22,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/hardware/ascend/ascend-optimization-casebook/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/hardware/ascend/ascend-optimization-casebook/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/hardware/ascend/ascend-optimization-casebook/SKILL.md",
         "src/inferrsi/data/skill_library/hardware/ascend/ascend-optimization-casebook/metadata.json",
@@ -40,7 +40,7 @@
   "source_files": {
     "src/inferrsi/data/skill_library/hardware/ascend/ascend-optimization-casebook/SKILL.md": "647f5e9a288f2c4e1617ba53df59d9d827df749eb4fe86182d21a9c41175caf8",
     "src/inferrsi/data/skill_library/hardware/ascend/ascend-optimization-casebook/metadata.json": "f425f7c1abbd616387273e8932c6bd02f4b1da012f099286ad4171e0f6037b22",
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190"
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
   },
   "skill_metadata": {
     "name": "ascend-optimization-casebook",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/ascend-performance-analysis.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/ascend-performance-analysis.json
index 9ee76ce80..9cd850847 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/ascend-performance-analysis.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/ascend-performance-analysis.json
@@ -22,8 +22,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/hardware/ascend/ascend-performance-analysis/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/hardware/ascend/ascend-performance-analysis/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/hardware/ascend/ascend-performance-analysis/SKILL.md",
         "src/inferrsi/data/skill_library/hardware/ascend/ascend-performance-analysis/metadata.json",
@@ -44,7 +44,7 @@
     "src/inferrsi/data/skill_library/hardware/ascend/ascend-performance-analysis/metadata.json": "9e705625c8308013c287576b0247da6f6116d817a4a535cbecff80abd6ea35ee",
     "src/inferrsi/data/skill_library/hardware/ascend/ascend-performance-analysis/references/cannbot-CANN-2.0-LICENSE.md": "89b1c114df62cfd183004f38d18d07f186436c3b37203dd69c3e84becf9d2282",
     "src/inferrsi/data/skill_library/hardware/ascend/ascend-performance-analysis/references/cannbot-stage-evidence.md": "f56e4125325c9c1b9235a1d64fd732861c7686444492422daff44ba8dd648be4",
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190"
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
   },
   "skill_metadata": {
     "name": "ascend-performance-analysis",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/ascend-video-diffusion-optimization.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/ascend-video-diffusion-optimization.json
index 1fff0e1b4..1c1bd3f9b 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/ascend-video-diffusion-optimization.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/ascend-video-diffusion-optimization.json
@@ -26,8 +26,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/hardware/ascend/ascend-video-diffusion-optimization/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/hardware/ascend/ascend-video-diffusion-optimization/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/hardware/ascend/ascend-video-diffusion-optimization/SKILL.md",
         "src/inferrsi/data/skill_library/hardware/ascend/ascend-video-diffusion-optimization/metadata.json",
@@ -46,7 +46,7 @@
     "src/inferrsi/data/skill_library/hardware/ascend/ascend-video-diffusion-optimization/SKILL.md": "53c21e64f0dd47f9510faef1f2c01e1283899787ba7ebfe70cdabd8e57a76f00",
     "src/inferrsi/data/skill_library/hardware/ascend/ascend-video-diffusion-optimization/metadata.json": "8c2b364298330751fd1252516159b16c27e90354e2a724782d2851e68e43cdb3",
     "src/inferrsi/data/skill_library/hardware/ascend/ascend-video-diffusion-optimization/references/field-notes-video-diffusion.md": "dadac7b23462fdc325218c72ddf8a2ac553151761a5d3e45b854b53d2c8c6908",
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190"
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
   },
   "skill_metadata": {
     "name": "ascend-video-diffusion-optimization",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/ascendc-tiling-design.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/ascendc-tiling-design.json
index f0e3b61fc..eb8d1a7cd 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/ascendc-tiling-design.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/ascendc-tiling-design.json
@@ -24,8 +24,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/hardware/ascend/ascendc-tiling-design/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/hardware/ascend/ascendc-tiling-design/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/hardware/ascend/ascendc-tiling-design/SKILL.md",
         "src/inferrsi/data/skill_library/hardware/ascend/ascendc-tiling-design/metadata.json",
@@ -42,7 +42,7 @@
   "source_files": {
     "src/inferrsi/data/skill_library/hardware/ascend/ascendc-tiling-design/SKILL.md": "93aa1b775ced304cebfe6c4dec0503ddb2260ecb3cb01edc17ce329446c5ce2b",
     "src/inferrsi/data/skill_library/hardware/ascend/ascendc-tiling-design/metadata.json": "f6d94f104cb70969024b2f4542078a240e200712c66279eca8b94f60bee1dd71",
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190"
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
   },
   "skill_metadata": {
     "name": "ascendc-tiling-design",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/cann-attribution-overlap.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/cann-attribution-overlap.json
index 2f08258b7..9b0132b55 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/cann-attribution-overlap.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/cann-attribution-overlap.json
@@ -24,8 +24,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/hardware/ascend/cann-attribution-overlap/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/hardware/ascend/cann-attribution-overlap/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/hardware/ascend/cann-attribution-overlap/SKILL.md",
         "src/inferrsi/data/skill_library/hardware/ascend/cann-attribution-overlap/metadata.json",
@@ -42,7 +42,7 @@
   "source_files": {
     "src/inferrsi/data/skill_library/hardware/ascend/cann-attribution-overlap/SKILL.md": "0e6f094dc554abea12e07def50f20b233b47127779a8aefad595b4ab1efd24c2",
     "src/inferrsi/data/skill_library/hardware/ascend/cann-attribution-overlap/metadata.json": "acccc8eb86de0df6caf78a9b3c6db76a2a686adc4a5ba0029a4d62d50ff9a8f8",
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190"
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
   },
   "skill_metadata": {
     "name": "cann-attribution-overlap",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/torchnpu-device-memory-lifecycle.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/torchnpu-device-memory-lifecycle.json
index e322e322b..6eb316093 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/torchnpu-device-memory-lifecycle.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/torchnpu-device-memory-lifecycle.json
@@ -22,8 +22,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/hardware/ascend/torchnpu-device-memory-lifecycle/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/hardware/ascend/torchnpu-device-memory-lifecycle/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/hardware/ascend/torchnpu-device-memory-lifecycle/SKILL.md",
         "src/inferrsi/data/skill_library/hardware/ascend/torchnpu-device-memory-lifecycle/metadata.json",
@@ -42,7 +42,7 @@
     "src/inferrsi/data/skill_library/hardware/ascend/torchnpu-device-memory-lifecycle/SKILL.md": "8428df7157af024fa17cfe0b3818ed54bd9f468306dda46eb8d4f0edf4f3f230",
     "src/inferrsi/data/skill_library/hardware/ascend/torchnpu-device-memory-lifecycle/metadata.json": "c650dce88c07b636d68e063dbda4d319f6f7f3ecbcdd5e47f393cbfc3af7c71d",
     "src/inferrsi/data/skill_library/hardware/ascend/torchnpu-device-memory-lifecycle/references/source-notes.md": "80cba4d3a884e7ee2618326cf74b9f9c3a34e4d9cf443707f6bbe673f2d5ba65",
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190"
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
   },
   "skill_metadata": {
     "name": "torchnpu-device-memory-lifecycle",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/torchnpu-pinned-host-memory-lifecycle.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/torchnpu-pinned-host-memory-lifecycle.json
index 1189f5c1b..cb9ca2c00 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/torchnpu-pinned-host-memory-lifecycle.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/ascend/skills/torchnpu-pinned-host-memory-lifecycle.json
@@ -22,8 +22,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/hardware/ascend/torchnpu-pinned-host-memory-lifecycle/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/hardware/ascend/torchnpu-pinned-host-memory-lifecycle/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/hardware/ascend/torchnpu-pinned-host-memory-lifecycle/SKILL.md",
         "src/inferrsi/data/skill_library/hardware/ascend/torchnpu-pinned-host-memory-lifecycle/metadata.json",
@@ -42,7 +42,7 @@
     "src/inferrsi/data/skill_library/hardware/ascend/torchnpu-pinned-host-memory-lifecycle/SKILL.md": "23e93ea7c98f99743509fdc04139d78ed2e7123c63e80b366c6beaba7eb70d0d",
     "src/inferrsi/data/skill_library/hardware/ascend/torchnpu-pinned-host-memory-lifecycle/metadata.json": "97836a38465106d816a9ace894a7b277d6d330f9accade8f8493ce6f0fb57ef9",
     "src/inferrsi/data/skill_library/hardware/ascend/torchnpu-pinned-host-memory-lifecycle/references/source-notes.md": "435193590b4a2ba32e2e4c6fa9d7e8f889528d412cbbb48e35ec5433f7dab384",
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190"
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
   },
   "skill_metadata": {
     "name": "torchnpu-pinned-host-memory-lifecycle",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/assess-serving-preemption.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/assess-serving-preemption.json
index 503808901..f5e145cef 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/assess-serving-preemption.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/assess-serving-preemption.json
@@ -16,8 +16,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/methods/assess-serving-preemption/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/methods/assess-serving-preemption/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/methods/assess-serving-preemption/SKILL.md",
         "src/inferrsi/data/skill_library/methods/assess-serving-preemption/metadata.json",
@@ -36,7 +36,7 @@
     "src/inferrsi/data/skill_library/methods/assess-serving-preemption/SKILL.md": "0a5f48d9352c5bf460775ef47f9136829978811f4a16e84983e299782cae305d",
     "src/inferrsi/data/skill_library/methods/assess-serving-preemption/metadata.json": "77ce5ba16245cbd80d46d0abdf80aaf26fc0924fa2d2a5624ff430de68a1daf0",
     "src/inferrsi/data/skill_library/methods/assess-serving-preemption/references/source-notes.md": "f29fcecfba0490008a354c11b42403983fe165e56ff907ad61cc6040d2b6407f",
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190"
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
   },
   "skill_metadata": {
     "name": "assess-serving-preemption",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/assess-split-backward-schedule.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/assess-split-backward-schedule.json
index 40a701930..4a226cb88 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/assess-split-backward-schedule.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/assess-split-backward-schedule.json
@@ -16,8 +16,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/methods/assess-split-backward-schedule/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/methods/assess-split-backward-schedule/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/methods/assess-split-backward-schedule/SKILL.md",
         "src/inferrsi/data/skill_library/methods/assess-split-backward-schedule/metadata.json",
@@ -36,7 +36,7 @@
     "src/inferrsi/data/skill_library/methods/assess-split-backward-schedule/SKILL.md": "1e64f28ea71672fb53c56f6108b87864738c6d03d6dcd32eedaef9e0625e96e3",
     "src/inferrsi/data/skill_library/methods/assess-split-backward-schedule/metadata.json": "c6acae57d70190cea0e18c71c79127962181e58169e011df75d8635b3db4daa7",
     "src/inferrsi/data/skill_library/methods/assess-split-backward-schedule/references/source-notes.md": "61b1609f71c37b61d136b8c4d39f793de87cffb1cf31c8ec737a1c3fa6fb1b4e",
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190"
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
   },
   "skill_metadata": {
     "name": "assess-split-backward-schedule",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/audit-fixed-workload-claim.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/audit-fixed-workload-claim.json
index 164d027a2..c2eed02a9 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/audit-fixed-workload-claim.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/audit-fixed-workload-claim.json
@@ -20,8 +20,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/methods/audit-fixed-workload-claim/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/methods/audit-fixed-workload-claim/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/methods/audit-fixed-workload-claim/SKILL.md",
         "src/inferrsi/data/skill_library/methods/audit-fixed-workload-claim/metadata.json",
@@ -38,7 +38,7 @@
   "source_files": {
     "src/inferrsi/data/skill_library/methods/audit-fixed-workload-claim/SKILL.md": "2be89b1340a3deb978c8e1792d18aeecbe8f5de81d4daf542389f53826b8ebee",
     "src/inferrsi/data/skill_library/methods/audit-fixed-workload-claim/metadata.json": "81edd75fa16939807548d0944e9203cfd7c0cb17e63463872d96ca2c1273d1d8",
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190"
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
   },
   "skill_metadata": {
     "name": "audit-fixed-workload-claim",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/audit-h3-metal-fasth3-semantics.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/audit-h3-metal-fasth3-semantics.json
index 4f40c1f4b..652747ffb 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/audit-h3-metal-fasth3-semantics.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/audit-h3-metal-fasth3-semantics.json
@@ -29,8 +29,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/repositories/h3-metal/audit-h3-metal-fasth3-semantics/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/repositories/h3-metal/audit-h3-metal-fasth3-semantics/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/repositories/h3-metal/audit-h3-metal-fasth3-semantics/LICENSE",
         "src/inferrsi/data/skill_library/repositories/h3-metal/audit-h3-metal-fasth3-semantics/SKILL.md",
@@ -65,7 +65,7 @@
     "src/inferrsi/data/skill_library/repositories/h3-metal/audit-h3-metal-fasth3-semantics/references/source-notes.md": "f9430103644eacd9c55ace6f84d8587fc0bc063bfcd9cb60e6045b139a9fee05",
     "src/inferrsi/data/skill_library/repositories/h3-metal/audit-h3-metal-fasth3-semantics/references/troubleshooting.md": "ed552d87f9fb9fb2c56de64df61f3fc6e2333cb049116cd0736f784c8a327612",
     "src/inferrsi/data/skill_library/repositories/h3-metal/audit-h3-metal-fasth3-semantics/scripts/audit_fasth3.py": "05bfb9348cf88a76f4fba8bc8a600fe54246d83a98c1685bc4847ace28e2fbac",
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190"
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
   },
   "skill_metadata": {
     "name": "audit-h3-metal-fasth3-semantics",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/audit-h3-runtime-residency.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/audit-h3-runtime-residency.json
index 003ffda39..152c3bd32 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/audit-h3-runtime-residency.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/audit-h3-runtime-residency.json
@@ -30,8 +30,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/repositories/h3-c/audit-h3-runtime-residency/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/repositories/h3-c/audit-h3-runtime-residency/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/repositories/h3-c/audit-h3-runtime-residency/LICENSE",
         "src/inferrsi/data/skill_library/repositories/h3-c/audit-h3-runtime-residency/SKILL.md",
@@ -66,7 +66,7 @@
     "src/inferrsi/data/skill_library/repositories/h3-c/audit-h3-runtime-residency/references/source-identity.json": "730f8dd23648a3f8f47b9726cc613d893a79d1817d515eec609a9b06d8a9ff90",
     "src/inferrsi/data/skill_library/repositories/h3-c/audit-h3-runtime-residency/references/source-notes.md": "7304bf9369efc157dc7aa40686dd828cc14db45c49442e908b3d3cc4b329eb86",
     "src/inferrsi/data/skill_library/repositories/h3-c/audit-h3-runtime-residency/references/validation.md": "0754873a758e048480938314894e25b504ae0c269549312e7f8e6020d83fe294",
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190"
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
   },
   "skill_metadata": {
     "name": "audit-h3-runtime-residency",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/audit-llama-runtime-state.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/audit-llama-runtime-state.json
index b8d72f000..f4cf1fd48 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/audit-llama-runtime-state.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/audit-llama-runtime-state.json
@@ -23,8 +23,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/repositories/llama-cpp/audit-llama-runtime-state/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/repositories/llama-cpp/audit-llama-runtime-state/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/repositories/llama-cpp/audit-llama-runtime-state/SKILL.md",
         "src/inferrsi/data/skill_library/repositories/llama-cpp/audit-llama-runtime-state/metadata.json",
@@ -43,7 +43,7 @@
     "src/inferrsi/data/skill_library/repositories/llama-cpp/audit-llama-runtime-state/SKILL.md": "95e81b6bc096a954fa73139ea5bc001d1efa6c99747161da62f012b494e5c013",
     "src/inferrsi/data/skill_library/repositories/llama-cpp/audit-llama-runtime-state/metadata.json": "1d30b31dc8352fc6a8d7b8abbef8118c840e47e5b90795dae0c758bf9717278c",
     "src/inferrsi/data/skill_library/repositories/llama-cpp/audit-llama-runtime-state/references/source-notes.md": "3d12a401e9decb55a2f4f3b2bfa0d360ef5110f63a0469dfe285654e78054190",
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190"
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
   },
   "skill_metadata": {
     "name": "audit-llama-runtime-state",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/audit-recurrent-kv-finalization.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/audit-recurrent-kv-finalization.json
index 11c783b3a..b12090c45 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/audit-recurrent-kv-finalization.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/audit-recurrent-kv-finalization.json
@@ -26,8 +26,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/repositories/vllm-rlt/audit-recurrent-kv-finalization/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/repositories/vllm-rlt/audit-recurrent-kv-finalization/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/repositories/vllm-rlt/audit-recurrent-kv-finalization/LICENSE",
         "src/inferrsi/data/skill_library/repositories/vllm-rlt/audit-recurrent-kv-finalization/SKILL.md",
@@ -52,7 +52,7 @@
     "src/inferrsi/data/skill_library/repositories/vllm-rlt/audit-recurrent-kv-finalization/references/NOTICE.md": "cca794656897ecdc37f983f7f23c4cd12306dbe3c9cd2f54fece6340753ae4cf",
     "src/inferrsi/data/skill_library/repositories/vllm-rlt/audit-recurrent-kv-finalization/references/source-identity.json": "5e52db0a0bda4673b9aac46a3fc035b824780def72921fee6796edb2ed9fca65",
     "src/inferrsi/data/skill_library/repositories/vllm-rlt/audit-recurrent-kv-finalization/references/source-notes.md": "0956f59cd9a1fac27da53228dd60c51a07b87639a43c09bfab296c4aeeceaab9",
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190"
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
   },
   "skill_metadata": {
     "name": "audit-recurrent-kv-finalization",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/audit-sd-cpp-residency.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/audit-sd-cpp-residency.json
index f26f7af97..3ea837069 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/audit-sd-cpp-residency.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/audit-sd-cpp-residency.json
@@ -25,8 +25,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/repositories/stable-diffusion-cpp/audit-sd-cpp-residency/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/repositories/stable-diffusion-cpp/audit-sd-cpp-residency/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/repositories/stable-diffusion-cpp/audit-sd-cpp-residency/LICENSE",
         "src/inferrsi/data/skill_library/repositories/stable-diffusion-cpp/audit-sd-cpp-residency/SKILL.md",
@@ -55,7 +55,7 @@
     "src/inferrsi/data/skill_library/repositories/stable-diffusion-cpp/audit-sd-cpp-residency/references/source-identity.json": "af4830370b6eddc2d176a9ae14da4ab56a90b0fe048a01a89ee051df4365d8e9",
     "src/inferrsi/data/skill_library/repositories/stable-diffusion-cpp/audit-sd-cpp-residency/references/source-notes.md": "c493af8f6fd11ab7c9a37c9eec78cac31d57c04403a790035efbb4fff99f4669",
     "src/inferrsi/data/skill_library/repositories/stable-diffusion-cpp/audit-sd-cpp-residency/references/validation-boundaries.md": "3aaf79b36192bf8ff3e8f95ac90f2e769aa065dfac1fe70368824f3591849b8d",
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190"
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
   },
   "skill_metadata": {
     "name": "audit-sd-cpp-residency",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/audit-sglang-diffusion-lifetimes.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/audit-sglang-diffusion-lifetimes.json
index 4bee4ca8d..0a21fddf0 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/audit-sglang-diffusion-lifetimes.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/audit-sglang-diffusion-lifetimes.json
@@ -23,8 +23,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/repositories/sglang/audit-sglang-diffusion-lifetimes/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/repositories/sglang/audit-sglang-diffusion-lifetimes/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/repositories/sglang/audit-sglang-diffusion-lifetimes/LICENSE",
         "src/inferrsi/data/skill_library/repositories/sglang/audit-sglang-diffusion-lifetimes/SKILL.md",
@@ -45,7 +45,7 @@
     "src/inferrsi/data/skill_library/repositories/sglang/audit-sglang-diffusion-lifetimes/SKILL.md": "b0a4ba4751087126143ef86983798deb0ed00e24b6ff06bf6ef76a9df6e274f2",
     "src/inferrsi/data/skill_library/repositories/sglang/audit-sglang-diffusion-lifetimes/metadata.json": "6ed57dfb4f25ecb34b81f2188b055e05d172ba2581e6a3b605b814b2f733a4d6",
     "src/inferrsi/data/skill_library/repositories/sglang/audit-sglang-diffusion-lifetimes/references/source-notes.md": "d0fcfd1177ae2f351041c5b2abba7f5121cec133006d9e4dca8f74ddb567332d",
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190"
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
   },
   "skill_metadata": {
     "name": "audit-sglang-diffusion-lifetimes",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/audit-vllm-kv-lifetimes.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/audit-vllm-kv-lifetimes.json
index 9115d501a..cf6eedc70 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/audit-vllm-kv-lifetimes.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/audit-vllm-kv-lifetimes.json
@@ -21,8 +21,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/repositories/vllm/audit-vllm-kv-lifetimes/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/repositories/vllm/audit-vllm-kv-lifetimes/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/repositories/vllm/audit-vllm-kv-lifetimes/SKILL.md",
         "src/inferrsi/data/skill_library/repositories/vllm/audit-vllm-kv-lifetimes/metadata.json",
@@ -41,7 +41,7 @@
     "src/inferrsi/data/skill_library/repositories/vllm/audit-vllm-kv-lifetimes/SKILL.md": "4c3f27459df239f9f4fc971c559d4c9ddfbfebd70507a8a93be8eae59a30b69d",
     "src/inferrsi/data/skill_library/repositories/vllm/audit-vllm-kv-lifetimes/metadata.json": "b43fc3a7ccae59c664fcdf9ff77962edcc07c2a3df8b59324db551642bd23e23",
     "src/inferrsi/data/skill_library/repositories/vllm/audit-vllm-kv-lifetimes/references/source-notes.md": "3665b034594fc8cc767c408002f4e16f8803369242405027c7f37ce6c9ecebe9",
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190"
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
   },
   "skill_metadata": {
     "name": "audit-vllm-kv-lifetimes",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/bound-windowed-attention-replay.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/bound-windowed-attention-replay.json
index c847bc97f..cfc37350e 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/bound-windowed-attention-replay.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/bound-windowed-attention-replay.json
@@ -23,8 +23,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/repositories/deepinfer/bound-windowed-attention-replay/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/repositories/deepinfer/bound-windowed-attention-replay/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/repositories/deepinfer/bound-windowed-attention-replay/SKILL.md",
         "src/inferrsi/data/skill_library/repositories/deepinfer/bound-windowed-attention-replay/metadata.json",
@@ -45,7 +45,7 @@
     "src/inferrsi/data/skill_library/repositories/deepinfer/bound-windowed-attention-replay/metadata.json": "014be24d0e669f4acbc0717dc323eec784026cf185186619d5a19d6d84496131",
     "src/inferrsi/data/skill_library/repositories/deepinfer/bound-windowed-attention-replay/references/source-notes.md": "fd917bd5ac80bee9c9f4525f8e0300f6d67dc144e2a16632af42a67f3d53d098",
     "src/inferrsi/data/skill_library/repositories/deepinfer/bound-windowed-attention-replay/references/validation.md": "60e71e70c5519bcdf6c171eec2be2f342e8801bedd746d6005f11adc646dfaff",
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190"
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
   },
   "skill_metadata": {
     "name": "bound-windowed-attention-replay",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/budget-speculative-verification.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/budget-speculative-verification.json
index 336ce150b..50303a878 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/budget-speculative-verification.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/budget-speculative-verification.json
@@ -23,8 +23,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/repositories/deepinfer/budget-speculative-verification/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/repositories/deepinfer/budget-speculative-verification/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/repositories/deepinfer/budget-speculative-verification/SKILL.md",
         "src/inferrsi/data/skill_library/repositories/deepinfer/budget-speculative-verification/metadata.json",
@@ -45,7 +45,7 @@
     "src/inferrsi/data/skill_library/repositories/deepinfer/budget-speculative-verification/metadata.json": "4d68bf4c1767eaee12ade963bbd551d0e61d1bcd81a984bb02fd3aae46965919",
     "src/inferrsi/data/skill_library/repositories/deepinfer/budget-speculative-verification/references/source-notes.md": "6fa955022f16cd598f44fd6521901d1cb868b06cec545cabe2b82769a641b5d6",
     "src/inferrsi/data/skill_library/repositories/deepinfer/budget-speculative-verification/references/validation.md": "673d93f6700052974b340fab25b3438e0e6344960569ac9abec7d7720f662288",
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190"
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
   },
   "skill_metadata": {
     "name": "budget-speculative-verification",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/calibrate-serving-simulation.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/calibrate-serving-simulation.json
index 39fad1569..75ad9ff2a 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/calibrate-serving-simulation.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/calibrate-serving-simulation.json
@@ -16,8 +16,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/methods/calibrate-serving-simulation/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/methods/calibrate-serving-simulation/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/methods/calibrate-serving-simulation/SKILL.md",
         "src/inferrsi/data/skill_library/methods/calibrate-serving-simulation/metadata.json",
@@ -36,7 +36,7 @@
     "src/inferrsi/data/skill_library/methods/calibrate-serving-simulation/SKILL.md": "ac773c6f1c110a664dfca28a3ae9129b10c07614e60ca3640dea776220e7b352",
     "src/inferrsi/data/skill_library/methods/calibrate-serving-simulation/metadata.json": "261ce9f644ea6eebb90c830f41fb8b320d2b401456e5bd8d494095a44b8620d0",
     "src/inferrsi/data/skill_library/methods/calibrate-serving-simulation/references/source-notes.md": "9696d3c0ef96b6611787446bd6f96d4d6255ca8d107105de8c1d9af83edb9fa5",
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190"
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
   },
   "skill_metadata": {
     "name": "calibrate-serving-simulation",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/cgroup-resource-controls.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/cgroup-resource-controls.json
index 77e4d4602..e244f7550 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/cgroup-resource-controls.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/cgroup-resource-controls.json
@@ -25,8 +25,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/hardware/host/cgroup-resource-controls/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/hardware/host/cgroup-resource-controls/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/hardware/host/cgroup-resource-controls/SKILL.md",
         "src/inferrsi/data/skill_library/hardware/host/cgroup-resource-controls/metadata.json",
@@ -43,7 +43,7 @@
   "source_files": {
     "src/inferrsi/data/skill_library/hardware/host/cgroup-resource-controls/SKILL.md": "fd1f5a7e6a55714730044155f5b38c1d700383d4e16c3e91963bbfbe1c773af5",
     "src/inferrsi/data/skill_library/hardware/host/cgroup-resource-controls/metadata.json": "19e40763fedc20990651bb989c32a7693306c5c54af46c3d4be77fd4f87b6c03",
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190"
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
   },
   "skill_metadata": {
     "name": "cgroup-resource-controls",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/checkpoint-recovery-consistency.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/checkpoint-recovery-consistency.json
index 69317f2e3..d64cd5ecd 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/checkpoint-recovery-consistency.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/checkpoint-recovery-consistency.json
@@ -22,8 +22,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/methods/checkpoint-recovery-consistency/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/methods/checkpoint-recovery-consistency/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/methods/checkpoint-recovery-consistency/SKILL.md",
         "src/inferrsi/data/skill_library/methods/checkpoint-recovery-consistency/metadata.json",
@@ -44,7 +44,7 @@
     "src/inferrsi/data/skill_library/methods/checkpoint-recovery-consistency/metadata.json": "4dc8659806539b4784ea9348a01d4194ac70d0bb297a916f90ff6419efa5c328",
     "src/inferrsi/data/skill_library/methods/checkpoint-recovery-consistency/references/dlrover-persistence-replicas.md": "055f4357f79026fcbcc766bbe77db610376ec353623f8898f5d14a30ee69e7bc",
     "src/inferrsi/data/skill_library/methods/checkpoint-recovery-consistency/references/sources.md": "9425b1aab963a2c8d35f3db897bd15ee69699bf5b0d0c8425c3c8bba3e9a0233",
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190"
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
   },
   "skill_metadata": {
     "name": "checkpoint-recovery-consistency",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/choose-parallel-training-plan.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/choose-parallel-training-plan.json
index abccc41d0..a269c5a05 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/choose-parallel-training-plan.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/choose-parallel-training-plan.json
@@ -16,8 +16,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/methods/choose-parallel-training-plan/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/methods/choose-parallel-training-plan/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/methods/choose-parallel-training-plan/SKILL.md",
         "src/inferrsi/data/skill_library/methods/choose-parallel-training-plan/metadata.json",
@@ -36,7 +36,7 @@
     "src/inferrsi/data/skill_library/methods/choose-parallel-training-plan/SKILL.md": "25d04aed9782b2a8ff1186828becdc7fce93c4418039ceaef9b8dcc5ec9bcd24",
     "src/inferrsi/data/skill_library/methods/choose-parallel-training-plan/metadata.json": "c6acae57d70190cea0e18c71c79127962181e58169e011df75d8635b3db4daa7",
     "src/inferrsi/data/skill_library/methods/choose-parallel-training-plan/references/source-notes.md": "9f9184d46b19312138a6e805e439e43dc0594a0cc6b46675255a582dc946ca5b",
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190"
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
   },
   "skill_metadata": {
     "name": "choose-parallel-training-plan",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/cite-check-downstream-pr.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/cite-check-downstream-pr.json
index 1dd7ec767..52df4999f 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/cite-check-downstream-pr.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/cite-check-downstream-pr.json
@@ -16,8 +16,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/workflows/cite-check-downstream-pr/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/workflows/cite-check-downstream-pr/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/skill-index.json",
         "src/inferrsi/data/skill_library/workflows/cite-check-downstream-pr/SKILL.md"
@@ -31,7 +31,7 @@
   },
   "checkout_ref": "checkout:cite-check-downstream-pr",
   "source_files": {
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190",
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317",
     "src/inferrsi/data/skill_library/workflows/cite-check-downstream-pr/SKILL.md": "485932cdf3b1fba5dc69be73e75d38f86a7b79442d26e82a296c963a7598d226"
   },
   "skill_metadata": {
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/claim-contract.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/claim-contract.json
index 9ba516bd5..fd356308d 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/claim-contract.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/claim-contract.json
@@ -16,8 +16,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/workflows/claim-contract/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/workflows/claim-contract/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/skill-index.json",
         "src/inferrsi/data/skill_library/workflows/claim-contract/SKILL.md"
@@ -31,7 +31,7 @@
   },
   "checkout_ref": "checkout:claim-contract",
   "source_files": {
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190",
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317",
     "src/inferrsi/data/skill_library/workflows/claim-contract/SKILL.md": "3af1dbbe81d91cb05fc2ef15a2ae25d7592a08558e1a5535f49c52f33a488c55"
   },
   "skill_metadata": {
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/comfy-ui.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/comfy-ui.json
index 8cd6bce38..2abcb51b8 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/comfy-ui.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/comfy-ui.json
@@ -22,8 +22,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/repositories/comfy-ui/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/repositories/comfy-ui/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/repositories/comfy-ui/LICENSE",
         "src/inferrsi/data/skill_library/repositories/comfy-ui/SKILL.md",
@@ -50,7 +50,7 @@
     "src/inferrsi/data/skill_library/repositories/comfy-ui/references/model-residency.md": "014c68386e7cdfdb7b4f566836e8b2ff61bd5a5a506e7b4723f83fe80c5d4de7",
     "src/inferrsi/data/skill_library/repositories/comfy-ui/references/repo-provenance.md": "ca996ef515af42a3ce304747fa3f12a31c00533deaace5ba2ce21aefa0a62e2e",
     "src/inferrsi/data/skill_library/repositories/comfy-ui/references/validation-boundaries.md": "1913dedbeb558d872d6a0e015cc11d822910c3cb259a236db29fb60ad8a64263",
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190"
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
   },
   "skill_metadata": {
     "name": "comfy-ui",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/competitor-scan.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/competitor-scan.json
index f8a01e0d4..ea767e40c 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/competitor-scan.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/competitor-scan.json
@@ -16,8 +16,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/workflows/competitor-scan/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/workflows/competitor-scan/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/skill-index.json",
         "src/inferrsi/data/skill_library/workflows/competitor-scan/SKILL.md"
@@ -31,7 +31,7 @@
   },
   "checkout_ref": "checkout:competitor-scan",
   "source_files": {
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190",
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317",
     "src/inferrsi/data/skill_library/workflows/competitor-scan/SKILL.md": "0ede58ace65a9324a538b4816add1684f230ba536d4364fe5a5dd2c0bbed3776"
   },
   "skill_metadata": {
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/dcp-async-save-performance.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/dcp-async-save-performance.json
index 4a2a71a50..c9e20850d 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/dcp-async-save-performance.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/dcp-async-save-performance.json
@@ -23,8 +23,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/frameworks/pytorch/dcp-async-save-performance/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/frameworks/pytorch/dcp-async-save-performance/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/frameworks/pytorch/dcp-async-save-performance/SKILL.md",
         "src/inferrsi/data/skill_library/frameworks/pytorch/dcp-async-save-performance/metadata.json",
@@ -43,7 +43,7 @@
     "src/inferrsi/data/skill_library/frameworks/pytorch/dcp-async-save-performance/SKILL.md": "645139ee0c88ec865cad952e24cda35f4648d4b777b732c9b7d274b6c7830845",
     "src/inferrsi/data/skill_library/frameworks/pytorch/dcp-async-save-performance/metadata.json": "db7ad368394ccbfdeff6a16e3423c300d51305b1f95ff5c3936df91738a8cc35",
     "src/inferrsi/data/skill_library/frameworks/pytorch/dcp-async-save-performance/references/sources.md": "22afe4f4d77f441096cc657d55d763473bfcbea8fb531c8e493fe49ab84955f7",
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190"
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
   },
   "skill_metadata": {
     "name": "dcp-async-save-performance",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/dense-performance-evidence.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/dense-performance-evidence.json
index b550229b9..86c00244f 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/dense-performance-evidence.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/dense-performance-evidence.json
@@ -21,8 +21,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/methods/dense-performance-evidence/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/methods/dense-performance-evidence/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/methods/dense-performance-evidence/SKILL.md",
         "src/inferrsi/data/skill_library/methods/dense-performance-evidence/metadata.json",
@@ -39,7 +39,7 @@
   "source_files": {
     "src/inferrsi/data/skill_library/methods/dense-performance-evidence/SKILL.md": "53cc1867a13f54d1006d50517dea8848869ca50df3f0cd610b2424cc8d18c933",
     "src/inferrsi/data/skill_library/methods/dense-performance-evidence/metadata.json": "a0d1e3b17140295510a764045d1b222e19be3a3406e6d1e7dddaad49ff474490",
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190"
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
   },
   "skill_metadata": {
     "name": "dense-performance-evidence",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/diagnose-vllm-admission-pressure.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/diagnose-vllm-admission-pressure.json
index b03fb1c00..261b4dce2 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/diagnose-vllm-admission-pressure.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/diagnose-vllm-admission-pressure.json
@@ -21,8 +21,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/repositories/vllm/diagnose-vllm-admission-pressure/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/repositories/vllm/diagnose-vllm-admission-pressure/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/repositories/vllm/diagnose-vllm-admission-pressure/SKILL.md",
         "src/inferrsi/data/skill_library/repositories/vllm/diagnose-vllm-admission-pressure/metadata.json",
@@ -67,7 +67,7 @@
     "src/inferrsi/data/skill_library/repositories/vllm/diagnose-vllm-admission-pressure/references/papers-20261015/references/loongserve-readset.md": "f1a383e7bdbbd5a5a472d7f6e1292c8a17d4910ff736ca5a6aa422e1bee74f8c",
     "src/inferrsi/data/skill_library/repositories/vllm/diagnose-vllm-admission-pressure/references/papers-20261015/references/orca-readset.md": "a497ef23b7e60c0fbc15d95b7e07cd20547cfe56f37c24dfe6ed70d79ef7a3cf",
     "src/inferrsi/data/skill_library/repositories/vllm/diagnose-vllm-admission-pressure/references/source-notes.md": "84033fd514773f10c0bddd9914fabb980497ba2b308f226a922c3e1888413827",
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190"
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
   },
   "skill_metadata": {
     "name": "diagnose-vllm-admission-pressure",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/diffusers-invocation-state.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/diffusers-invocation-state.json
index 0927d8546..5fd806dd1 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/diffusers-invocation-state.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/diffusers-invocation-state.json
@@ -22,8 +22,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/frameworks/diffusers/diffusers-invocation-state/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/frameworks/diffusers/diffusers-invocation-state/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/frameworks/diffusers/diffusers-invocation-state/SKILL.md",
         "src/inferrsi/data/skill_library/frameworks/diffusers/diffusers-invocation-state/metadata.json",
@@ -46,7 +46,7 @@
     "src/inferrsi/data/skill_library/frameworks/diffusers/diffusers-invocation-state/references/ownership.md": "40688cc44183027efbf2fca8d3548c299cdf2e33fef54f76ab3549353eb9bd1b",
     "src/inferrsi/data/skill_library/frameworks/diffusers/diffusers-invocation-state/references/source-map.md": "10562a75683bd9bd07ec2e37a1cbd0f93f2534d6f85b55a566adb2d97485215f",
     "src/inferrsi/data/skill_library/frameworks/diffusers/diffusers-invocation-state/references/validation.md": "5a0b854b87d091d7d505b0bdd5c03330052507a851ab6edf35b93a6675f2d4f7",
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190"
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
   },
   "skill_metadata": {
     "name": "diffusers-invocation-state",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/draft-acceptance-issue.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/draft-acceptance-issue.json
index 8b35785db..a47f59b95 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/draft-acceptance-issue.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/draft-acceptance-issue.json
@@ -16,8 +16,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/workflows/draft-acceptance-issue/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/workflows/draft-acceptance-issue/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/skill-index.json",
         "src/inferrsi/data/skill_library/workflows/draft-acceptance-issue/SKILL.md"
@@ -31,7 +31,7 @@
   },
   "checkout_ref": "checkout:draft-acceptance-issue",
   "source_files": {
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190",
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317",
     "src/inferrsi/data/skill_library/workflows/draft-acceptance-issue/SKILL.md": "ce983acfb7d7d3272775e7cf00c92e4ddadd97511b602575a76e5ab7ade01b56"
   },
   "skill_metadata": {
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/dwarfstar.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/dwarfstar.json
index 88c601413..1cceae90d 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/dwarfstar.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/dwarfstar.json
@@ -25,8 +25,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/repositories/dwarfstar/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/repositories/dwarfstar/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/repositories/dwarfstar/LICENSE",
         "src/inferrsi/data/skill_library/repositories/dwarfstar/SKILL.md",
@@ -71,7 +71,7 @@
     "src/inferrsi/data/skill_library/repositories/dwarfstar/sub-skills/serving-state/SKILL.md": "71c27212847146376d72e6cc6d22abe2f7eed56402c5917cd7dfcda0b2aa5982",
     "src/inferrsi/data/skill_library/repositories/dwarfstar/sub-skills/serving-state/references/state-contracts.md": "8f74a07c947021e3069468268f6b5aa2548a17db518c5c0505b84dde73a1f31c",
     "src/inferrsi/data/skill_library/repositories/dwarfstar/sub-skills/serving-state/references/troubleshooting.md": "c9eebe4109d418c4dfb9bb2425c145ebc1a52da9b665e0556149bc31a442e2fc",
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190"
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
   },
   "skill_metadata": {
     "name": "dwarfstar",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/evaluate-elastic-serving-placement.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/evaluate-elastic-serving-placement.json
index 89e5768c3..7ce699705 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/evaluate-elastic-serving-placement.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/evaluate-elastic-serving-placement.json
@@ -16,8 +16,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/methods/evaluate-elastic-serving-placement/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/methods/evaluate-elastic-serving-placement/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/methods/evaluate-elastic-serving-placement/SKILL.md",
         "src/inferrsi/data/skill_library/methods/evaluate-elastic-serving-placement/metadata.json",
@@ -36,7 +36,7 @@
     "src/inferrsi/data/skill_library/methods/evaluate-elastic-serving-placement/SKILL.md": "82d0099d3c33706f46c657d4fa7f76d3d486e019b36760d1de001921328a60cc",
     "src/inferrsi/data/skill_library/methods/evaluate-elastic-serving-placement/metadata.json": "37379a4d4a058ca46da2b20a920dc01be9ec11e557c32b2332702ec76027f4b5",
     "src/inferrsi/data/skill_library/methods/evaluate-elastic-serving-placement/references/source-notes.md": "ea416831fd7d0874e06d0a76fb0c7196df53072b717673a0cdf635db6bfd917b",
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190"
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
   },
   "skill_metadata": {
     "name": "evaluate-elastic-serving-placement",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/generation-cache-lifecycle.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/generation-cache-lifecycle.json
index a480d0f76..abc16cb9a 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/generation-cache-lifecycle.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/generation-cache-lifecycle.json
@@ -22,8 +22,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/repositories/mlx-lm/generation-cache-lifecycle/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/repositories/mlx-lm/generation-cache-lifecycle/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/repositories/mlx-lm/generation-cache-lifecycle/SKILL.md",
         "src/inferrsi/data/skill_library/repositories/mlx-lm/generation-cache-lifecycle/metadata.json",
@@ -48,7 +48,7 @@
     "src/inferrsi/data/skill_library/repositories/mlx-lm/generation-cache-lifecycle/references/generation-batch.md": "f16d3c57d45e63bcaffd3871aa41c76972ee11f7dbf1e12b45d9b170303cad9b",
     "src/inferrsi/data/skill_library/repositories/mlx-lm/generation-cache-lifecycle/references/source-map.md": "91d26181f246307f2113f94b12ac9587d689b654dcd216e9f14b11cf5ebb650c",
     "src/inferrsi/data/skill_library/repositories/mlx-lm/generation-cache-lifecycle/references/validation.md": "73b5dd347017baba519ecc85415aead9993c1751d116b9ba4f07689f58c5f783",
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190"
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
   },
   "skill_metadata": {
     "name": "generation-cache-lifecycle",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/host-memory-management.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/host-memory-management.json
index 1a2d57d6f..7076497e8 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/host-memory-management.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/host-memory-management.json
@@ -27,8 +27,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/hardware/host/host-memory-management/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/hardware/host/host-memory-management/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/hardware/host/host-memory-management/SKILL.md",
         "src/inferrsi/data/skill_library/hardware/host/host-memory-management/metadata.json",
@@ -51,7 +51,7 @@
     "src/inferrsi/data/skill_library/hardware/host/host-memory-management/references/vllm-ascend-LICENSE.md": "c71d239df91726fc519c6eb72d318ec65820627232b2f796219e87dcf35d0ab4",
     "src/inferrsi/data/skill_library/hardware/host/host-memory-management/references/vllm-ascend-native-offload-readset.json": "cc6c2011be746002ba3ff96ccefab1809863748ddd17eb102b75e7b0e134fe55",
     "src/inferrsi/data/skill_library/hardware/host/host-memory-management/references/vllm-ascend-native-offload.md": "6c2a72b5ba3708a01d229e0576570837a617e6a42dbb0b5312a26a65a6c31542",
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190"
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
   },
   "skill_metadata": {
     "name": "host-memory-management",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/host-overhead-optimization.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/host-overhead-optimization.json
index 33b9bb470..cd198e542 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/host-overhead-optimization.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/host-overhead-optimization.json
@@ -22,8 +22,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/hardware/host/host-overhead-optimization/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/hardware/host/host-overhead-optimization/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/hardware/host/host-overhead-optimization/SKILL.md",
         "src/inferrsi/data/skill_library/hardware/host/host-overhead-optimization/metadata.json",
@@ -42,7 +42,7 @@
     "src/inferrsi/data/skill_library/hardware/host/host-overhead-optimization/SKILL.md": "36ef88826a32ac3147b869001004ded17a8d5a491e25bfe105005ba39ea9d1eb",
     "src/inferrsi/data/skill_library/hardware/host/host-overhead-optimization/metadata.json": "ae71c389dfdb1a19d267feecb4c6f1fbb4c9afad501ff6d8fab7e38b29bcc37a",
     "src/inferrsi/data/skill_library/hardware/host/host-overhead-optimization/references/deepinfer-zero-overhead-scheduling.md": "f452bf40132c9195d739f1c86b3aa18857138c7ed0f1861af5853d93b9955924",
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190"
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
   },
   "skill_metadata": {
     "name": "host-overhead-optimization",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/husky.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/husky.json
index f4badb2ec..79394bbc9 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/husky.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/husky.json
@@ -21,8 +21,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/repositories/husky/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/repositories/husky/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/repositories/husky/SKILL.md",
         "src/inferrsi/data/skill_library/repositories/husky/metadata.json",
@@ -51,7 +51,7 @@
     "src/inferrsi/data/skill_library/repositories/husky/references/troubleshooting.md": "3948711437d25822d389dd857c72dc7fcc15727c9048d2e3eae3317d97aca6dd",
     "src/inferrsi/data/skill_library/repositories/husky/references/verification-protocol.md": "83c5a966ed3f721b412b8d8417af8336c5a8c1e46044a888720936bf2f1deeab",
     "src/inferrsi/data/skill_library/repositories/husky/scripts/audit_evidence_card.py": "291e62701ef7a37a5bfab963ea4e20abb94245bc015584223a101f18b1490895",
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190"
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
   },
   "skill_metadata": {
     "name": "husky",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/inference-benchmark-design.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/inference-benchmark-design.json
index 8434cb260..e6f1719f1 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/inference-benchmark-design.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/inference-benchmark-design.json
@@ -30,8 +30,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/methods/inference-benchmark-design/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/methods/inference-benchmark-design/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/methods/inference-benchmark-design/SKILL.md",
         "src/inferrsi/data/skill_library/methods/inference-benchmark-design/metadata.json",
@@ -52,7 +52,7 @@
     "src/inferrsi/data/skill_library/methods/inference-benchmark-design/metadata.json": "18310402a8ade03fd48ac687f79ba9d447871e90337de84c0e1bc15c0b827306",
     "src/inferrsi/data/skill_library/methods/inference-benchmark-design/references/benchmark-commands.md": "0898424e0bff3a6246e1eeeda6abd6c5bd974c047fe32f3bbd5ab9d31ce17cd3",
     "src/inferrsi/data/skill_library/methods/inference-benchmark-design/references/task-benchmark-selection.md": "0d4bb080a754e26846e81680023db0dff5e6a4c96be4327c411c3b72ac4ef8e2",
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190"
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
   },
   "skill_metadata": {
     "name": "inference-benchmark-design",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/inference-optimization-casebook.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/inference-optimization-casebook.json
index e67cdaa03..00062f481 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/inference-optimization-casebook.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/inference-optimization-casebook.json
@@ -21,8 +21,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/methods/inference-optimization-casebook/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/methods/inference-optimization-casebook/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/methods/inference-optimization-casebook/SKILL.md",
         "src/inferrsi/data/skill_library/methods/inference-optimization-casebook/cases/admission-drain-occupancy.json",
@@ -55,7 +55,7 @@
     "src/inferrsi/data/skill_library/methods/inference-optimization-casebook/cases/qwen-b300-rejected-directions.json": "146ea1564a46ed03092c7dc5796b318fc05b04b1821105ee00eb7fa7fa1333e2",
     "src/inferrsi/data/skill_library/methods/inference-optimization-casebook/cases/qwen-b300-rejected-directions.md": "cce2930f544a0f78f21797fba4d8ded298e4a81f76c34d7977b0e31d361bc738",
     "src/inferrsi/data/skill_library/methods/inference-optimization-casebook/metadata.json": "fb55067fb45970319787068a3206110641cceaf0ccbabd706be6d0b3d2afc2ed",
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190"
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
   },
   "skill_metadata": {
     "name": "inference-optimization-casebook",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/kernel-writing-guide.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/kernel-writing-guide.json
index ae41bd4ae..0ba450924 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/kernel-writing-guide.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/kernel-writing-guide.json
@@ -39,8 +39,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/methods/kernel-writing-guide/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/methods/kernel-writing-guide/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/methods/kernel-writing-guide/SKILL.md",
         "src/inferrsi/data/skill_library/methods/kernel-writing-guide/cases/causal-conv1d-dispatch-state.md",
@@ -89,7 +89,7 @@
     "src/inferrsi/data/skill_library/methods/kernel-writing-guide/references/sglang-gdn-causal-conv-contract.md": "c9f27b0231ddec3d14d6fa0831d1494ffd8e11bd8ad704e04be3bc4a4ffc0e58",
     "src/inferrsi/data/skill_library/methods/kernel-writing-guide/references/sglang-gdn-source-files.json": "4c7ec2cfd2c29297e185ce09b4d9b2040fbe3635943c112f9ebb71dca8ab381b",
     "src/inferrsi/data/skill_library/methods/kernel-writing-guide/references/vllm-gdn-causal-conv-contract.md": "eec09a6e95983c46b67d5f066d34419c17819899f5d8e3c65f2f163f2ada0fde",
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190"
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
   },
   "skill_metadata": {
     "name": "kernel-writing-guide",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/kv-tier-management.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/kv-tier-management.json
index 35bb3f433..92d5a40ee 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/kv-tier-management.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/kv-tier-management.json
@@ -24,8 +24,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/repositories/lmcache/kv-tier-management/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/repositories/lmcache/kv-tier-management/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/repositories/lmcache/kv-tier-management/SKILL.md",
         "src/inferrsi/data/skill_library/repositories/lmcache/kv-tier-management/metadata.json",
@@ -42,7 +42,7 @@
   "source_files": {
     "src/inferrsi/data/skill_library/repositories/lmcache/kv-tier-management/SKILL.md": "e4b78f368d70c8abb74a935acb910414015f1aeec1b50f7a295e836822ec4b27",
     "src/inferrsi/data/skill_library/repositories/lmcache/kv-tier-management/metadata.json": "c221189b8d85fac3bd41ef1c7cd560f1aebc8cfa0381c182d1d571a7169be829",
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190"
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
   },
   "skill_metadata": {
     "name": "kv-tier-management",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/lazy-evaluation-stream-lifetimes.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/lazy-evaluation-stream-lifetimes.json
index 22adfd1eb..19ed2be0c 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/lazy-evaluation-stream-lifetimes.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/lazy-evaluation-stream-lifetimes.json
@@ -22,8 +22,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/frameworks/mlx/lazy-evaluation-stream-lifetimes/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/frameworks/mlx/lazy-evaluation-stream-lifetimes/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/frameworks/mlx/lazy-evaluation-stream-lifetimes/SKILL.md",
         "src/inferrsi/data/skill_library/frameworks/mlx/lazy-evaluation-stream-lifetimes/metadata.json",
@@ -46,7 +46,7 @@
     "src/inferrsi/data/skill_library/frameworks/mlx/lazy-evaluation-stream-lifetimes/references/execution-lifetime.md": "c4bc41e08aca68f661325f61e6f0db1760a079dc362f295c847bdc374fee3bf6",
     "src/inferrsi/data/skill_library/frameworks/mlx/lazy-evaluation-stream-lifetimes/references/source-map.md": "798b9cb89ed6b885eb47464ca9f137d2802a54571b9327e631d7742d3e5abd5d",
     "src/inferrsi/data/skill_library/frameworks/mlx/lazy-evaluation-stream-lifetimes/references/validation.md": "ebfc00bf7d6ab0a1dc415953ae84d56ccdc713d5614aff85e433c4a843793bff",
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190"
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
   },
   "skill_metadata": {
     "name": "lazy-evaluation-stream-lifetimes",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/lily.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/lily.json
index 03ad4419d..f5923fa44 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/lily.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/lily.json
@@ -25,8 +25,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/repositories/lily/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/repositories/lily/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/repositories/lily/LICENSE",
         "src/inferrsi/data/skill_library/repositories/lily/NOTICE.md",
@@ -77,7 +77,7 @@
     "src/inferrsi/data/skill_library/repositories/lily/sub-skills/serving-cache/references/api-and-session.md": "7262bb077c24f2f7549d450473ec7944829940d09cca89db9c12f37ac90d5db0",
     "src/inferrsi/data/skill_library/repositories/lily/sub-skills/serving-cache/references/troubleshooting.md": "6e2f36d2f7ab9eae665d87a676e34c0ebd2f46f3df6fa5f2e589a306749a87b5",
     "src/inferrsi/data/skill_library/repositories/lily/sub-skills/serving-cache/scripts/request_template.py": "d6b00c9f0b38c7fb0a0e52e396993a832542ea0aa6e2456026fabc0c39a4d33a",
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190"
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
   },
   "skill_metadata": {
     "name": "lily",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/lithos-metal.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/lithos-metal.json
new file mode 100644
index 000000000..a0161c74f
--- /dev/null
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/lithos-metal.json
@@ -0,0 +1,172 @@
+{
+  "schema_version": 1,
+  "id": "inferrsi.common.skills.lithos_metal",
+  "title": "lithos-metal",
+  "repo": "inferrsi",
+  "tier": "contribution",
+  "backend": "common",
+  "component": "skills",
+  "scenario": "skills.discovery",
+  "status": "imported",
+  "summary": "Inspect Lithos Metal compilation and fusion, GPU-resident generation and DSpark state, or local serving and exact-prefix caches on Apple silicon; use when changing recipes, diagnosing stale state, or qualifying an inference claim.",
+  "tags": [
+    "lithos-metal",
+    "metal",
+    "apple-silicon",
+    "repository-guide"
+  ],
+  "origin": [
+    "external"
+  ],
+  "sources": [
+    {
+      "type": "git",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/repositories/lithos-metal/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
+      "paths": [
+        "src/inferrsi/data/skill_library/repositories/lithos-metal/LICENSE",
+        "src/inferrsi/data/skill_library/repositories/lithos-metal/SKILL.md",
+        "src/inferrsi/data/skill_library/repositories/lithos-metal/metadata.json",
+        "src/inferrsi/data/skill_library/repositories/lithos-metal/references/mlx-license.md",
+        "src/inferrsi/data/skill_library/repositories/lithos-metal/references/notices.md",
+        "src/inferrsi/data/skill_library/repositories/lithos-metal/references/package-api.md",
+        "src/inferrsi/data/skill_library/repositories/lithos-metal/references/repo-provenance.md",
+        "src/inferrsi/data/skill_library/repositories/lithos-metal/references/source-notes.md",
+        "src/inferrsi/data/skill_library/repositories/lithos-metal/references/troubleshooting.md",
+        "src/inferrsi/data/skill_library/repositories/lithos-metal/references/upstream-notice.md",
+        "src/inferrsi/data/skill_library/repositories/lithos-metal/scripts/inspect_contract.py",
+        "src/inferrsi/data/skill_library/repositories/lithos-metal/sub-skills/compile-fusion/SKILL.md",
+        "src/inferrsi/data/skill_library/repositories/lithos-metal/sub-skills/compile-fusion/metadata.json",
+        "src/inferrsi/data/skill_library/repositories/lithos-metal/sub-skills/compile-fusion/references/mechanisms.md",
+        "src/inferrsi/data/skill_library/repositories/lithos-metal/sub-skills/compile-fusion/references/troubleshooting.md",
+        "src/inferrsi/data/skill_library/repositories/lithos-metal/sub-skills/generation-state/SKILL.md",
+        "src/inferrsi/data/skill_library/repositories/lithos-metal/sub-skills/generation-state/metadata.json",
+        "src/inferrsi/data/skill_library/repositories/lithos-metal/sub-skills/generation-state/references/state-machine.md",
+        "src/inferrsi/data/skill_library/repositories/lithos-metal/sub-skills/generation-state/references/troubleshooting.md",
+        "src/inferrsi/data/skill_library/repositories/lithos-metal/sub-skills/serving-cache/SKILL.md",
+        "src/inferrsi/data/skill_library/repositories/lithos-metal/sub-skills/serving-cache/metadata.json",
+        "src/inferrsi/data/skill_library/repositories/lithos-metal/sub-skills/serving-cache/references/troubleshooting.md",
+        "src/inferrsi/data/skill_library/repositories/lithos-metal/sub-skills/serving-cache/references/workflows.md",
+        "src/inferrsi/data/skill_library/skill-index.json"
+      ],
+      "content_sha256": "4c71ba2f26c275829466bba2412738aa27799842e642c75cdfd36d9ed8713f7f"
+    }
+  ],
+  "index_type": "packaged-skill",
+  "skill_documents": {
+    "original": "data/skill_library/repositories/lithos-metal/SKILL.md"
+  },
+  "checkout_ref": "checkout:lithos-metal",
+  "source_files": {
+    "src/inferrsi/data/skill_library/repositories/lithos-metal/LICENSE": "cfc7749b96f63bd31c3c42b5c471bf756814053e847c10f3eb003417bc523d30",
+    "src/inferrsi/data/skill_library/repositories/lithos-metal/SKILL.md": "4c71ba2f26c275829466bba2412738aa27799842e642c75cdfd36d9ed8713f7f",
+    "src/inferrsi/data/skill_library/repositories/lithos-metal/metadata.json": "ea2d4e73e33a49433a2834d19489653f828e56f69c4975cc32cf508176eb24ae",
+    "src/inferrsi/data/skill_library/repositories/lithos-metal/references/mlx-license.md": "ccfab7ccb2ea306f71531c8ca77bb55507606cd90768b1e32b8b52ab5b48cf01",
+    "src/inferrsi/data/skill_library/repositories/lithos-metal/references/notices.md": "b2c9d2481ef2c9d16b3240c951ae65db417e5d309cd06b7917db9d84fe175886",
+    "src/inferrsi/data/skill_library/repositories/lithos-metal/references/package-api.md": "4f3ecfabafb2a15eddf21ccf964b8c37635bbb7862418a2c404e7049c0fa2898",
+    "src/inferrsi/data/skill_library/repositories/lithos-metal/references/repo-provenance.md": "c0c7c3dd45eab0d55bcf154760156db58e8e365f9cfb2c6bd246dd6a06f3ef3c",
+    "src/inferrsi/data/skill_library/repositories/lithos-metal/references/source-notes.md": "110dbc995c7fc727632017d4c26c71df7ecf12d4df5d89df0cd4afd75c5892f2",
+    "src/inferrsi/data/skill_library/repositories/lithos-metal/references/troubleshooting.md": "1894bbe1a18018f81318b11e2150201ee97f3819f84a7a81f3cf5fb01549c165",
+    "src/inferrsi/data/skill_library/repositories/lithos-metal/references/upstream-notice.md": "cd80bddd6b6204dbc61c6ac762ed206c73d9264755d9884383d0860de29d185d",
+    "src/inferrsi/data/skill_library/repositories/lithos-metal/scripts/inspect_contract.py": "f8365f87af2135fcf0a2c8baaefac563c30989282fe080a02d958a37a215515c",
+    "src/inferrsi/data/skill_library/repositories/lithos-metal/sub-skills/compile-fusion/SKILL.md": "e0d927763f3774440abddefc45913528946cd29f7f089766fab47b212f9b6693",
+    "src/inferrsi/data/skill_library/repositories/lithos-metal/sub-skills/compile-fusion/metadata.json": "b3ff3702531e023b49beded950151ec1256aab4fbf4c02e9212ef0202d3ad72e",
+    "src/inferrsi/data/skill_library/repositories/lithos-metal/sub-skills/compile-fusion/references/mechanisms.md": "6173a8b73ec5138cb12791f108f3d5d83278a117d443d48187c842e16bdc6eff",
+    "src/inferrsi/data/skill_library/repositories/lithos-metal/sub-skills/compile-fusion/references/troubleshooting.md": "f18ca9324d9a80f2ba327a1ad4ccf84a01049080366fe12410f2800d5f178232",
+    "src/inferrsi/data/skill_library/repositories/lithos-metal/sub-skills/generation-state/SKILL.md": "bd8cd27ddb18a04c9e2b31ce9f16e676864133e160267473979de828b1915f63",
+    "src/inferrsi/data/skill_library/repositories/lithos-metal/sub-skills/generation-state/metadata.json": "dacba1953fad82b32b98f76bd48009b04109f1ae446cc68be522b02a64306157",
+    "src/inferrsi/data/skill_library/repositories/lithos-metal/sub-skills/generation-state/references/state-machine.md": "d812ad31d4e343539119cf82f8ce241fe14c691dd73398048893cc5de3ab8e22",
+    "src/inferrsi/data/skill_library/repositories/lithos-metal/sub-skills/generation-state/references/troubleshooting.md": "7ca629e66ef3b0a67077a7d2536befef683315ad01939728a9e6e7523cc8414c",
+    "src/inferrsi/data/skill_library/repositories/lithos-metal/sub-skills/serving-cache/SKILL.md": "01c38c2678d6e8292a3908bf5619b332432c5f1289fd5d5a67eb44a2141adb90",
+    "src/inferrsi/data/skill_library/repositories/lithos-metal/sub-skills/serving-cache/metadata.json": "2fca222efac7752e29f236f40a0bc38275045826b55955687e39c64ffb69f58f",
+    "src/inferrsi/data/skill_library/repositories/lithos-metal/sub-skills/serving-cache/references/troubleshooting.md": "52c6e27a76547475b5c66121f873f4d2c1662c811a69da64ce85614f1bd44749",
+    "src/inferrsi/data/skill_library/repositories/lithos-metal/sub-skills/serving-cache/references/workflows.md": "298d27d3c748fe22774b872ed0465b707e53812379a7a538c95f25d1727fc796",
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
+  },
+  "skill_metadata": {
+    "name": "lithos-metal",
+    "description": "Inspect Lithos Metal compilation and fusion, GPU-resident generation and DSpark state, or local serving and exact-prefix caches on Apple silicon; use when changing recipes, diagnosing stale state, or qualifying an inference claim.",
+    "profile": "local-skill-v1",
+    "schema_version": 1,
+    "kind": "skill",
+    "origin": [
+      "external"
+    ],
+    "status": "candidate",
+    "empirical_validation": "not_run",
+    "tags": [
+      "lithos-metal",
+      "metal",
+      "apple-silicon",
+      "repository-guide"
+    ],
+    "sources": [
+      {
+        "type": "git",
+        "url": "https://github.com/lithos-ai/lithos-metal",
+        "commit": "6653b7df5f2c4392ec7b038960f2796b1499923f",
+        "paths": [
+          "kernels/common/spec_ops.metal",
+          "monolith/backends/metal/base.py",
+          "monolith/backends/metal/registry.py",
+          "monolith/compiler/arena.py",
+          "monolith/compiler/barriers.py",
+          "monolith/compiler/coverage.py",
+          "monolith/compiler/decoder_fusion.py",
+          "monolith/compiler/emit.py",
+          "monolith/compiler/region_fusion.py",
+          "monolith/compiler/static_fusion.py",
+          "monolith/core/step_state.py",
+          "monolith/generate.py",
+          "monolith/runtime/engine.py",
+          "monolith/runtime/prefix_cache.py",
+          "monolith/runtime/prefix_store.py",
+          "monolith/serve.py",
+          "monolith/serving/cache.py",
+          "monolith/serving/lifecycle.py",
+          "monolith/serving/setup.py",
+          "monolith/spec/dspark/model.py",
+          "runtime/src/metal_core.mm"
+        ],
+        "readset": "references/source-notes.md"
+      }
+    ],
+    "applicability": {
+      "scope": "Pinned Lithos Metal 0.1.2 source/API contracts; Apple silicon Metal runtime; model and recipe qualification required separately."
+    },
+    "limitations": [
+      "Source-derived candidate; empirical efficacy validation not_run.",
+      "CPU/software and synthetic Metal contracts do not qualify full-model/DSpark numerics, quality or E2E performance.",
+      "No model downloads, live DisCo deployment or inference backend integration is provided."
+    ],
+    "provenance_readsets": [
+      "references/source-notes.md"
+    ],
+    "consumption_mode": "packaged",
+    "license": "Apache-2.0",
+    "disable-model-invocation": true,
+    "metadata": {
+      "disco-role": "operating"
+    }
+  },
+  "applicability": {
+    "scope": "Pinned Lithos Metal 0.1.2 source/API contracts; Apple silicon Metal runtime; model and recipe qualification required separately."
+  },
+  "recommendations": [
+    "Read the pinned SKILL.md and its prerequisites before choosing this procedure.",
+    "Read the packaged references and owner-local cases alongside the procedure."
+  ],
+  "limitations": [
+    "This generated abstract points to the complete packaged skill; external evidence and upstream repositories remain source links.",
+    "The imported status describes this index, not the original skill; its declared status is candidate.",
+    "Reading or indexing does not authorize execution, validate efficacy, promote a candidate or grant a license.",
+    "The backend categorizes the procedure's hardware dependency; it does not certify compatibility, and skills.discovery is a discovery scenario.",
+    "Source-derived candidate; empirical efficacy validation not_run.",
+    "CPU/software and synthetic Metal contracts do not qualify full-model/DSpark numerics, quality or E2E performance.",
+    "No model downloads, live DisCo deployment or inference backend integration is provided."
+  ],
+  "verification": {
+    "locally_verified": false,
+    "evidence_level": "guidance"
+  }
+}
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/local-dream.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/local-dream.json
index bb870451e..dd74195d9 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/local-dream.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/local-dream.json
@@ -29,8 +29,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/repositories/local-dream/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/repositories/local-dream/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/repositories/local-dream/SKILL.md",
         "src/inferrsi/data/skill_library/repositories/local-dream/metadata.json",
@@ -61,7 +61,7 @@
     "src/inferrsi/data/skill_library/repositories/local-dream/references/source-identity.json": "b85f028bb96a37852e9d9780b5d5052fdd9c8f3a704ad391b6381adb58291347",
     "src/inferrsi/data/skill_library/repositories/local-dream/references/upstream-license.md": "41003d4a74749c0220e33dd415042164b5a1093ed401f36277234f772d22d3d0",
     "src/inferrsi/data/skill_library/repositories/local-dream/scripts/screen_declarations.py": "ccd7d0d55be9a47484ec3a74498d8edef8aba55287f92287775c8d82050c8932",
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190"
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
   },
   "skill_metadata": {
     "name": "local-dream",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/mistral-rs.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/mistral-rs.json
index dbe3c7035..26ff396ec 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/mistral-rs.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/mistral-rs.json
@@ -26,8 +26,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/repositories/mistral-rs/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/repositories/mistral-rs/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/repositories/mistral-rs/LICENSE",
         "src/inferrsi/data/skill_library/repositories/mistral-rs/SKILL.md",
@@ -90,7 +90,7 @@
     "src/inferrsi/data/skill_library/repositories/mistral-rs/sub-skills/qualification/references/troubleshooting.md": "d933bea6e9e493faa8cd0f4103ea33197ace7f7f901359f8cf0d6bab2edcdb4d",
     "src/inferrsi/data/skill_library/repositories/mistral-rs/sub-skills/qualification/references/warmup-inventory.md": "1e7af7d63c6853ede4232b5a12cf68aba6597c4a305015366f1c2c917e3caf01",
     "src/inferrsi/data/skill_library/repositories/mistral-rs/sub-skills/qualification/scripts/warmup_shape_inventory.py": "6aa920ca1cb534dec41f5dd2d833c3e760b234ab2c6439b6c4250b8a30b74055",
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190"
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
   },
   "skill_metadata": {
     "name": "mistral-rs",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/mlc-llm.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/mlc-llm.json
index 1c9aa73fb..8fc7f6dcc 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/mlc-llm.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/mlc-llm.json
@@ -27,8 +27,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/repositories/mlc-llm/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/repositories/mlc-llm/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/repositories/mlc-llm/LICENSE",
         "src/inferrsi/data/skill_library/repositories/mlc-llm/SKILL.md",
@@ -97,7 +97,7 @@
     "src/inferrsi/data/skill_library/repositories/mlc-llm/sub-skills/runtime-kv-limits/references/limits-and-workspace.md": "77105db95de2b8ea27ed6953b5441c8af061fb745ddb13d3904f8773fc21e879",
     "src/inferrsi/data/skill_library/repositories/mlc-llm/sub-skills/runtime-kv-limits/references/troubleshooting.md": "b89255fc04507185bc8bcf74de779730b009b12d13bb38ad0193205dfed8e921",
     "src/inferrsi/data/skill_library/repositories/mlc-llm/sub-skills/runtime-kv-limits/scripts/audit_kv_limit_declarations.py": "d82a6dda5f5931eb369003a5e1f419bfbdd82304766a18be8ea1e1849e26b77d",
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190"
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
   },
   "skill_metadata": {
     "name": "mlc-llm",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/mlx-swift-lm.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/mlx-swift-lm.json
index f84eca50e..04f8beaaa 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/mlx-swift-lm.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/mlx-swift-lm.json
@@ -27,8 +27,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/repositories/mlx-swift-lm/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/repositories/mlx-swift-lm/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/repositories/mlx-swift-lm/LICENSE",
         "src/inferrsi/data/skill_library/repositories/mlx-swift-lm/SKILL.md",
@@ -79,7 +79,7 @@
     "src/inferrsi/data/skill_library/repositories/mlx-swift-lm/sub-skills/generation-lifecycle/references/ownership-and-termination.md": "56fdec7ebc89b5c217c7e5c203886bef1d3c2eb5794852897bdb5941ab1db2a0",
     "src/inferrsi/data/skill_library/repositories/mlx-swift-lm/sub-skills/generation-lifecycle/references/troubleshooting.md": "de28b8ca9cf95be2a4e5762f938acfadb6d4b53459dbf86df4b2b2ec64ec347f",
     "src/inferrsi/data/skill_library/repositories/mlx-swift-lm/sub-skills/generation-lifecycle/scripts/audit_generation_evidence.py": "2aba6caf42ca1b3fc55299fd96419af8681c37750c21a7ba75205a43baa9c7b1",
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190"
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
   },
   "skill_metadata": {
     "name": "mlx-swift-lm",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/npu-skill-factory.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/npu-skill-factory.json
index a205e602b..bb24d6465 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/npu-skill-factory.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/npu-skill-factory.json
@@ -22,8 +22,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/workflows/npu-skill-factory/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/workflows/npu-skill-factory/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/skill-index.json",
         "src/inferrsi/data/skill_library/workflows/npu-skill-factory/SKILL.md",
@@ -54,7 +54,7 @@
   },
   "checkout_ref": "checkout:npu-skill-factory",
   "source_files": {
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190",
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317",
     "src/inferrsi/data/skill_library/workflows/npu-skill-factory/SKILL.md": "74cd863df2e80dc2bda1713d24207809173a3116f30325e3487e1e3517cfec1e",
     "src/inferrsi/data/skill_library/workflows/npu-skill-factory/metadata.json": "12fc9272bd5fbf02dd9b897d8ca40984b72ae54f20492ace5304711bbc4d71c3",
     "src/inferrsi/data/skill_library/workflows/npu-skill-factory/references/papers-20261015/COMPARISON.md": "38587678a96581f5c2dc697d982127343565504bffa191352811dffcc165aa23",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/optimization-search-policy.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/optimization-search-policy.json
index 34b0b9198..8094cecb3 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/optimization-search-policy.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/optimization-search-policy.json
@@ -21,8 +21,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/methods/optimization-search-policy/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/methods/optimization-search-policy/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/methods/optimization-search-policy/SKILL.md",
         "src/inferrsi/data/skill_library/methods/optimization-search-policy/metadata.json",
@@ -51,7 +51,7 @@
     "src/inferrsi/data/skill_library/methods/optimization-search-policy/references/papers-20261015/knowledge/adaptation-interfaces.md": "5e761957cb5ecd93926e6ffc13deee7c830c6e8f0ee3c3846e9dd3b50dc74446",
     "src/inferrsi/data/skill_library/methods/optimization-search-policy/references/papers-20261015/references/source-notes.md": "54d83c3713045d59a48821bcf93598374ba4a630e3891ad39498f919317c4fb8",
     "src/inferrsi/data/skill_library/methods/optimization-search-policy/references/papers-20261015/updates/optimization-search-policy.md": "d3db44fa9a6831b86a02c638b323ed943e69fe256866213f10d68c72d762879e",
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190"
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
   },
   "skill_metadata": {
     "name": "optimization-search-policy",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/os-process-placement.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/os-process-placement.json
index 6478e787b..4007d717c 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/os-process-placement.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/os-process-placement.json
@@ -24,8 +24,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/hardware/host/os-process-placement/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/hardware/host/os-process-placement/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/hardware/host/os-process-placement/SKILL.md",
         "src/inferrsi/data/skill_library/hardware/host/os-process-placement/metadata.json",
@@ -44,7 +44,7 @@
     "src/inferrsi/data/skill_library/hardware/host/os-process-placement/SKILL.md": "2003492e4d89fde0443b180d74933e764bd1979435e109c8d26e20eb8f6e5053",
     "src/inferrsi/data/skill_library/hardware/host/os-process-placement/metadata.json": "ec02995c075f0a540921f57b4846f2cafe334eb90f44389f17c3c50457c3e9c5",
     "src/inferrsi/data/skill_library/hardware/host/os-process-placement/references/llama-numa-thread-readset.md": "748b254b3b043a5ddb0f480eb4a566854d27eedd7b111499863e9bf0a5a6951e",
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190"
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
   },
   "skill_metadata": {
     "name": "os-process-placement",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/performance-ceiling-analysis.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/performance-ceiling-analysis.json
index 05cd55eff..b99f400c1 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/performance-ceiling-analysis.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/performance-ceiling-analysis.json
@@ -22,8 +22,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/methods/performance-ceiling-analysis/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/methods/performance-ceiling-analysis/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/methods/performance-ceiling-analysis/SKILL.md",
         "src/inferrsi/data/skill_library/methods/performance-ceiling-analysis/metadata.json",
@@ -78,7 +78,7 @@
     "src/inferrsi/data/skill_library/methods/performance-ceiling-analysis/references/roofline-contract.md": "419ec4bcaec45ac579a133bce083d4f48779f26e664d7a86f720fa65f343b1d1",
     "src/inferrsi/data/skill_library/methods/performance-ceiling-analysis/references/roofline-example.json": "597cd3205df043cff3ae9d15924c77934e8d8eb300c00c4840f6bb2921f90faf",
     "src/inferrsi/data/skill_library/methods/performance-ceiling-analysis/scripts/roofline.py": "524de48a6764401695fe1444f1fa3d38f363069b0cd2fba4715a86ca4b28a5d4",
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190"
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
   },
   "skill_metadata": {
     "name": "performance-ceiling-analysis",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/plan-activation-residency.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/plan-activation-residency.json
index ce3b38abe..222ca7ca1 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/plan-activation-residency.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/plan-activation-residency.json
@@ -16,8 +16,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/methods/plan-activation-residency/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/methods/plan-activation-residency/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/methods/plan-activation-residency/SKILL.md",
         "src/inferrsi/data/skill_library/methods/plan-activation-residency/metadata.json",
@@ -36,7 +36,7 @@
     "src/inferrsi/data/skill_library/methods/plan-activation-residency/SKILL.md": "542d8a1e6963f5019986e1fbd79d3a247a045a51487c09313026087ac457c15e",
     "src/inferrsi/data/skill_library/methods/plan-activation-residency/metadata.json": "c6acae57d70190cea0e18c71c79127962181e58169e011df75d8635b3db4daa7",
     "src/inferrsi/data/skill_library/methods/plan-activation-residency/references/source-notes.md": "5ff52655cfd06fcf970ed2a7d4d4fe1c31115b102af2d41497de0c74848ed2b3",
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190"
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
   },
   "skill_metadata": {
     "name": "plan-activation-residency",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/project-maya.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/project-maya.json
index e9887e8cc..0e5f9fd2b 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/project-maya.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/project-maya.json
@@ -25,8 +25,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/repositories/project-maya/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/repositories/project-maya/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/repositories/project-maya/LICENSE",
         "src/inferrsi/data/skill_library/repositories/project-maya/SKILL.md",
@@ -83,7 +83,7 @@
     "src/inferrsi/data/skill_library/repositories/project-maya/sub-skills/serving-state/references/native-state.md": "e5a7ee5e777998db0fb5422cf5b1ab18dbd67b4cbbfcfd6628ed6e40d8709c61",
     "src/inferrsi/data/skill_library/repositories/project-maya/sub-skills/serving-state/references/troubleshooting.md": "97dea7b2d26fe165d2f64a1a63c91d099dddb2d3b95c16b667f5173b3ffa9d2d",
     "src/inferrsi/data/skill_library/repositories/project-maya/sub-skills/serving-state/scripts/request_template.py": "d91b03fbbd7003012a92bd7259d39604b9a91ee5707da2d3366e7c0fb47edd6a",
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190"
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
   },
   "skill_metadata": {
     "name": "project-maya",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/repository-evolution-analysis.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/repository-evolution-analysis.json
index 3dbabf30f..aa985d99b 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/repository-evolution-analysis.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/repository-evolution-analysis.json
@@ -21,8 +21,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/methods/repository-evolution-analysis/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/methods/repository-evolution-analysis/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/methods/repository-evolution-analysis/SKILL.md",
         "src/inferrsi/data/skill_library/methods/repository-evolution-analysis/metadata.json",
@@ -43,7 +43,7 @@
     "src/inferrsi/data/skill_library/methods/repository-evolution-analysis/metadata.json": "98ef753a0e74c46f59a69ae267bdc0a5d293761e6f27edc1822fa8bd4548c4d6",
     "src/inferrsi/data/skill_library/methods/repository-evolution-analysis/references/capability-presence-audit.md": "c21eb4a6aabe811a49a6540a34c54dbcfdcf7efae3052a79b54034c7838026a5",
     "src/inferrsi/data/skill_library/methods/repository-evolution-analysis/references/rtk-consumer-path-transition.md": "a70b1d128fb5e0ac12292a290d624f218a0831d7581fe8980b62661473e0f2b7",
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190"
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
   },
   "skill_metadata": {
     "name": "repository-evolution-analysis",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/rsi-evidence.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/rsi-evidence.json
index 5bec52bfe..6f53f4db6 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/rsi-evidence.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/rsi-evidence.json
@@ -16,8 +16,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/workflows/rsi-evidence/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/workflows/rsi-evidence/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/skill-index.json",
         "src/inferrsi/data/skill_library/workflows/rsi-evidence/SKILL.md",
@@ -38,7 +38,7 @@
   },
   "checkout_ref": "checkout:rsi-evidence",
   "source_files": {
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190",
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317",
     "src/inferrsi/data/skill_library/workflows/rsi-evidence/SKILL.md": "08f76e12f6f8e8893f03bcdc8152a96c0c99442e7faecbc2a67a18c632fcb243",
     "src/inferrsi/data/skill_library/workflows/rsi-evidence/references/loopx-goal-governance/LICENSE": "cfc7749b96f63bd31c3c42b5c471bf756814053e847c10f3eb003417bc523d30",
     "src/inferrsi/data/skill_library/workflows/rsi-evidence/references/loopx-goal-governance/LICENSE-MIT.md": "1df1cf519b9e5d8acbbc72820299d7aa95afa1acbfc1f43a15109ad1d59975f3",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/skill-evolution-playbook.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/skill-evolution-playbook.json
index 59abe8a8a..aee0e55a1 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/skill-evolution-playbook.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/skill-evolution-playbook.json
@@ -21,8 +21,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/methods/skill-evolution-playbook/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/methods/skill-evolution-playbook/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/methods/skill-evolution-playbook/SKILL.md",
         "src/inferrsi/data/skill_library/methods/skill-evolution-playbook/metadata.json",
@@ -39,7 +39,7 @@
   "source_files": {
     "src/inferrsi/data/skill_library/methods/skill-evolution-playbook/SKILL.md": "9cd37f21fe0d862069e4798edca8a04d59f8b840e55d521901594a63dddd9ba3",
     "src/inferrsi/data/skill_library/methods/skill-evolution-playbook/metadata.json": "92b36338b9257e2e41b95e4370694d7de52ed84bc7971056b3bfced07b0993d8",
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190"
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
   },
   "skill_metadata": {
     "name": "skill-evolution-playbook",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/splash.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/splash.json
index c6a1aef0b..5d82ddc5a 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/splash.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/splash.json
@@ -25,8 +25,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/repositories/splash/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/repositories/splash/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/repositories/splash/LICENSE",
         "src/inferrsi/data/skill_library/repositories/splash/SKILL.md",
@@ -77,7 +77,7 @@
     "src/inferrsi/data/skill_library/repositories/splash/sub-skills/serving-cache/SKILL.md": "c16f911498e116ff9b31b5a6f0ddd3ae6517b51c0e9d079c337f36a5bcac1df8",
     "src/inferrsi/data/skill_library/repositories/splash/sub-skills/serving-cache/references/state-and-api.md": "592603177473ac8e024b7dc19a3d3faedd7c1e0312137eab8bf9a62d2f567be6",
     "src/inferrsi/data/skill_library/repositories/splash/sub-skills/serving-cache/references/troubleshooting.md": "4fda461042a3de43c109d7a3cec67542c7d61fe68412bf91606220f2e13e576e",
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190"
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
   },
   "skill_metadata": {
     "name": "splash",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/storage-io-measurement.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/storage-io-measurement.json
index 845f5ea83..ecb0a4ce8 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/storage-io-measurement.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/storage-io-measurement.json
@@ -24,8 +24,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/hardware/host/storage-io-measurement/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/hardware/host/storage-io-measurement/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/hardware/host/storage-io-measurement/SKILL.md",
         "src/inferrsi/data/skill_library/hardware/host/storage-io-measurement/metadata.json",
@@ -42,7 +42,7 @@
   "source_files": {
     "src/inferrsi/data/skill_library/hardware/host/storage-io-measurement/SKILL.md": "63b883c10a8d3e7d918abfc7175b72e2b401ff1e02697f94b8382ec1c8f369b7",
     "src/inferrsi/data/skill_library/hardware/host/storage-io-measurement/metadata.json": "0a41c7b00978332b54ee1e4a7c1ade88ecde3e961128f24d1b4f7708caff5ff5",
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190"
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
   },
   "skill_metadata": {
     "name": "storage-io-measurement",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/training-input-pipeline.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/training-input-pipeline.json
index d9f7c8a1d..66d66befb 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/training-input-pipeline.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/training-input-pipeline.json
@@ -23,8 +23,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/methods/training-input-pipeline/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/methods/training-input-pipeline/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/methods/training-input-pipeline/SKILL.md",
         "src/inferrsi/data/skill_library/methods/training-input-pipeline/metadata.json",
@@ -43,7 +43,7 @@
     "src/inferrsi/data/skill_library/methods/training-input-pipeline/SKILL.md": "64bd2c6b795d966045d1d3cca55976d3d49b08b06df7c7e06a70c2824cf55b7e",
     "src/inferrsi/data/skill_library/methods/training-input-pipeline/metadata.json": "80659099fce6a6ec05e9084b75452c877a5812e30f60a78f31ef8c7efcdc1bac",
     "src/inferrsi/data/skill_library/methods/training-input-pipeline/references/sources.md": "60e0f3e88655d8e1d5eb4797f39ca4370deda01614f265b8d2ea4559d8ac020f",
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190"
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
   },
   "skill_metadata": {
     "name": "training-input-pipeline",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/training-resource-budget.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/training-resource-budget.json
index dd24dad96..dea87856a 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/training-resource-budget.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/training-resource-budget.json
@@ -23,8 +23,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/methods/training-resource-budget/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/methods/training-resource-budget/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/methods/training-resource-budget/SKILL.md",
         "src/inferrsi/data/skill_library/methods/training-resource-budget/metadata.json",
@@ -57,7 +57,7 @@
     "src/inferrsi/data/skill_library/methods/training-resource-budget/references/papers-20261015/references/source-readset.md": "2a5c166767f56fa47e7eba7386f1f36da8cb73702e5b5bbe14d6a456ca9ebfae",
     "src/inferrsi/data/skill_library/methods/training-resource-budget/references/sources.md": "6d248f58af8b350fc66c22898988d6f451ac60d03bcd805cd6e84df5e55174ac",
     "src/inferrsi/data/skill_library/methods/training-resource-budget/references/stas-decision-notes.md": "f1a993e6c4ca9a035a2effcaf5500081f227b0cc56cc885a6b22fda0041fc744",
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190"
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
   },
   "skill_metadata": {
     "name": "training-resource-budget",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/video-diffusion-optimization.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/video-diffusion-optimization.json
index 8f2c64231..c3dffbed8 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/video-diffusion-optimization.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/video-diffusion-optimization.json
@@ -26,8 +26,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/methods/video-diffusion-optimization/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/methods/video-diffusion-optimization/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/methods/video-diffusion-optimization/SKILL.md",
         "src/inferrsi/data/skill_library/methods/video-diffusion-optimization/assets/campaign.template.json",
@@ -48,7 +48,7 @@
     "src/inferrsi/data/skill_library/methods/video-diffusion-optimization/assets/campaign.template.json": "1a37f1d136612143ddab8f0cbbf424c80433b4da5312edb18d07c0b9d87ef657",
     "src/inferrsi/data/skill_library/methods/video-diffusion-optimization/metadata.json": "b27d69142dfea5dd2820170475c2bf6cced2d5eefd1c43b3413c41849af0fd02",
     "src/inferrsi/data/skill_library/methods/video-diffusion-optimization/scripts/evaluate_target.py": "9e84df051e03d567746fd73ae7243efb2f3c5ec806fef824d3c4d42a8eb9704c",
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190"
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
   },
   "skill_metadata": {
     "name": "video-diffusion-optimization",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/vllm-metal.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/vllm-metal.json
index f80cb682a..d98cc79ce 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/vllm-metal.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/vllm-metal.json
@@ -26,8 +26,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/repositories/vllm-metal/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/repositories/vllm-metal/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/repositories/vllm-metal/LICENSE",
         "src/inferrsi/data/skill_library/repositories/vllm-metal/SKILL.md",
@@ -86,7 +86,7 @@
     "src/inferrsi/data/skill_library/repositories/vllm-metal/sub-skills/qualification/references/speculation-and-pipeline.md": "64faced7f31d489e9b0c4b9523a803ff1a76dca0f9315bdc6d0b83935bd338a4",
     "src/inferrsi/data/skill_library/repositories/vllm-metal/sub-skills/qualification/references/troubleshooting.md": "a2e85696120ffc2af4249ac9c1e1b96320e02287e5099577c0474c025ab4f5f6",
     "src/inferrsi/data/skill_library/repositories/vllm-metal/sub-skills/qualification/scripts/summarize_serving_records.py": "78e660a13a6a6941d006cb11fc2f401a34f489e54273860ae6a208dca6a6a3c2",
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190"
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
   },
   "skill_metadata": {
     "name": "vllm-metal",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/weight-loading-optimization.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/weight-loading-optimization.json
index 44f49460d..661f5c0c4 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/weight-loading-optimization.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/common/skills/weight-loading-optimization.json
@@ -29,8 +29,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/methods/weight-loading-optimization/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/methods/weight-loading-optimization/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/methods/weight-loading-optimization/SKILL.md",
         "src/inferrsi/data/skill_library/methods/weight-loading-optimization/metadata.json",
@@ -53,7 +53,7 @@
     "src/inferrsi/data/skill_library/methods/weight-loading-optimization/references/fastsafetensors-transfer-readset.md": "5a39f14b102b0cee304541e011aa31e44e6b13527d1bee9987c3f3b66c064d3d",
     "src/inferrsi/data/skill_library/methods/weight-loading-optimization/references/flare-preparation-boundary.md": "1d7922785a196bdbafdc9791fe79428465a5acc0292ca8f99446f8c764b35454",
     "src/inferrsi/data/skill_library/methods/weight-loading-optimization/references/llama-loading-readset.md": "6c640f28ee3f0253a17b81461c865cb9f947367002cb75bf72550c4572d02e97",
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190"
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
   },
   "skill_metadata": {
     "name": "weight-loading-optimization",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/cuda/skills/audit-afd-p2p-exchanges.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/cuda/skills/audit-afd-p2p-exchanges.json
index 172bb9da3..c53c6cc1d 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/cuda/skills/audit-afd-p2p-exchanges.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/cuda/skills/audit-afd-p2p-exchanges.json
@@ -23,8 +23,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/repositories/afd-plugin/audit-afd-p2p-exchanges/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/repositories/afd-plugin/audit-afd-p2p-exchanges/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/repositories/afd-plugin/audit-afd-p2p-exchanges/SKILL.md",
         "src/inferrsi/data/skill_library/repositories/afd-plugin/audit-afd-p2p-exchanges/metadata.json",
@@ -45,7 +45,7 @@
     "src/inferrsi/data/skill_library/repositories/afd-plugin/audit-afd-p2p-exchanges/metadata.json": "4e6100dfb9fc7bc00b19915f4fd38ad509704f2a609c658adf4d632d4e05b6c1",
     "src/inferrsi/data/skill_library/repositories/afd-plugin/audit-afd-p2p-exchanges/references/afd-plugin-LICENSE.md": "0e8566ea362276878e1f38f97da49964dbea301c15d4aa9d727f97a4c39fd943",
     "src/inferrsi/data/skill_library/repositories/afd-plugin/audit-afd-p2p-exchanges/references/source-notes.md": "5073d87450cb60d1b245308dff8807e73749c07de976d50f2fa26694fad81a72",
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190"
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
   },
   "skill_metadata": {
     "name": "audit-afd-p2p-exchanges",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/cuda/skills/audit-flare-session-lifecycle.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/cuda/skills/audit-flare-session-lifecycle.json
index 2180b1d43..a78f89860 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/cuda/skills/audit-flare-session-lifecycle.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/cuda/skills/audit-flare-session-lifecycle.json
@@ -29,8 +29,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/repositories/flare-local-inference/audit-flare-session-lifecycle/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/repositories/flare-local-inference/audit-flare-session-lifecycle/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/repositories/flare-local-inference/audit-flare-session-lifecycle/LICENSE",
         "src/inferrsi/data/skill_library/repositories/flare-local-inference/audit-flare-session-lifecycle/SKILL.md",
@@ -57,7 +57,7 @@
     "src/inferrsi/data/skill_library/repositories/flare-local-inference/audit-flare-session-lifecycle/references/lifecycle.md": "107ebe3c22e5dba8007c3b8ea61149f10bc55c51b557a4230f3e5de859bb0f7c",
     "src/inferrsi/data/skill_library/repositories/flare-local-inference/audit-flare-session-lifecycle/references/source-ledger.json": "61e51a97ffd7c1fa7866bdd440b024d0c5e0065635ac8d0ba53d2c69c7f9e470",
     "src/inferrsi/data/skill_library/repositories/flare-local-inference/audit-flare-session-lifecycle/references/source-notes.md": "fc7ceb819b91ef1731d8b7783f543c92ef6bb5c97ef5f897c055824f8a560168",
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190"
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
   },
   "skill_metadata": {
     "name": "audit-flare-session-lifecycle",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/cuda/skills/diagnose-attention-bottlenecks.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/cuda/skills/diagnose-attention-bottlenecks.json
index a605915a0..1259a655f 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/cuda/skills/diagnose-attention-bottlenecks.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/cuda/skills/diagnose-attention-bottlenecks.json
@@ -16,8 +16,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/hardware/cuda/diagnose-attention-bottlenecks/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/hardware/cuda/diagnose-attention-bottlenecks/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/hardware/cuda/diagnose-attention-bottlenecks/SKILL.md",
         "src/inferrsi/data/skill_library/hardware/cuda/diagnose-attention-bottlenecks/metadata.json",
@@ -52,7 +52,7 @@
     "src/inferrsi/data/skill_library/hardware/cuda/diagnose-attention-bottlenecks/references/papers-20261015/cases/partitioning-after-fusion.md": "4d4cabe0991eab554dcfbd4fa9c8199e61fa4767709b0c4cd646e2588bb398a2",
     "src/inferrsi/data/skill_library/hardware/cuda/diagnose-attention-bottlenecks/references/papers-20261015/knowledge/attention-cost-model-boundaries.md": "5ed8f888443e38ccb72545b6025447aab7b92936f78c151fed680e582126239b",
     "src/inferrsi/data/skill_library/hardware/cuda/diagnose-attention-bottlenecks/references/sources.md": "53436e0de4e6dda2d29426591bf61f8526378cdc6f9ec7967b88b1bcd4c61e6f",
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190"
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
   },
   "skill_metadata": {
     "name": "diagnose-attention-bottlenecks",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/cuda/skills/host-overhead-profiling.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/cuda/skills/host-overhead-profiling.json
index 2106f26a6..5b43491a3 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/cuda/skills/host-overhead-profiling.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/cuda/skills/host-overhead-profiling.json
@@ -22,8 +22,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/hardware/cuda/host-overhead-profiling/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/hardware/cuda/host-overhead-profiling/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/hardware/cuda/host-overhead-profiling/SKILL.md",
         "src/inferrsi/data/skill_library/hardware/cuda/host-overhead-profiling/metadata.json",
@@ -40,7 +40,7 @@
   "source_files": {
     "src/inferrsi/data/skill_library/hardware/cuda/host-overhead-profiling/SKILL.md": "326e566fbe30d3ba04e4b50c3e87aad9df037f0ac4ad6fe129165ae97dc5f91d",
     "src/inferrsi/data/skill_library/hardware/cuda/host-overhead-profiling/metadata.json": "76e34c1c64aba473a9e126b032a776417caa0f913f6ac3cf4dba3a5d123b1ee0",
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190"
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
   },
   "skill_metadata": {
     "name": "host-overhead-profiling",
diff --git a/src/inferrsi/data/repo_packs/inferrsi/knowledge/cuda/skills/reason-hopper-kernels.json b/src/inferrsi/data/repo_packs/inferrsi/knowledge/cuda/skills/reason-hopper-kernels.json
index f95b7e16d..186a290a3 100644
--- a/src/inferrsi/data/repo_packs/inferrsi/knowledge/cuda/skills/reason-hopper-kernels.json
+++ b/src/inferrsi/data/repo_packs/inferrsi/knowledge/cuda/skills/reason-hopper-kernels.json
@@ -36,8 +36,8 @@
   "sources": [
     {
       "type": "git",
-      "url": "https://github.com/InferMatrixLab/InferRSI/blob/2f72a0ccabd87eefdc76223c9b82f7091c379da1/src/inferrsi/data/skill_library/hardware/cuda/reason-hopper-kernels/SKILL.md",
-      "commit": "2f72a0ccabd87eefdc76223c9b82f7091c379da1",
+      "url": "https://github.com/InferMatrixLab/InferRSI/blob/26743abcfae48cd6223bd757e74f193d97e3e5a7/src/inferrsi/data/skill_library/hardware/cuda/reason-hopper-kernels/SKILL.md",
+      "commit": "26743abcfae48cd6223bd757e74f193d97e3e5a7",
       "paths": [
         "src/inferrsi/data/skill_library/hardware/cuda/reason-hopper-kernels/SKILL.md",
         "src/inferrsi/data/skill_library/hardware/cuda/reason-hopper-kernels/metadata.json",
@@ -72,7 +72,7 @@
     "src/inferrsi/data/skill_library/hardware/cuda/reason-hopper-kernels/references/source-inventory.json": "1dc09c895c970a0c9708158bcc6dd65cdb736274fd2c35d44542c70407d3849d",
     "src/inferrsi/data/skill_library/hardware/cuda/reason-hopper-kernels/references/sources.md": "6b74b9e774f2a80db0860510d9e6940d36284430f29c1e124f115a9eaf25833d",
     "src/inferrsi/data/skill_library/hardware/cuda/reason-hopper-kernels/references/synchronization.md": "4322f6b3c0e4368bf8c192b4ec752529767f50d19c0c38dad8fdf9cd781e41d9",
-    "src/inferrsi/data/skill_library/skill-index.json": "531190c468b6fb385b9b4aa71e666e8dd20d92aa5adc54bf4040f2096e3dd190"
+    "src/inferrsi/data/skill_library/skill-index.json": "b5a9f05bd8179644f3fea79d7eb04fe4a8baad9e8d68aa8d0f1db213cf633317"
   },
   "skill_metadata": {
     "name": "reason-hopper-kernels",
diff --git a/src/inferrsi/data/skill_library/README.md b/src/inferrsi/data/skill_library/README.md
index 729affa56..4e4b21967 100644
--- a/src/inferrsi/data/skill_library/README.md
+++ b/src/inferrsi/data/skill_library/README.md
@@ -1,7 +1,7 @@
 # InferRSI skill library
 
-The English-only library contains **120 skill documents**: 76 registered roots,
-21 separately indexed engine/router documents and 23 AREX graph sub-skills.
+The English-only library contains **124 skill documents**: 77 registered roots,
+21 separately indexed engine/router documents and 26 AREX graph sub-skills.
 Complete procedures, metadata, references, cases and source notices ship with the package.
 
 ## Browse by owner
@@ -20,7 +20,7 @@ existing resource directory.
 
 ## Read registered skills
 
-[`skill-index.json`](skill-index.json) maps 76 stable skill IDs to their canonical
+[`skill-index.json`](skill-index.json) maps 77 stable skill IDs to their canonical
 paths and backend scopes. Install the optional `skills` extra for these commands:
 
 ```bash
@@ -137,12 +137,13 @@ scheduling/placement proposals remain separate from both implementation pins.
 
 Five repository entry skills, 15 focused sub-skills and one research router
 provide source-grounded procedures. These 21 documents are available through
-the links and catalog records below, separately from the 76-ID registry.
+the links and catalog records below, separately from the 77-ID registry.
 The registered ComfyUI and stable-diffusion.cpp audits are listed alongside
 these imports and use generated abstracts rather than runtime repository packs.
 
 | Repository | Focus | Entry | Catalog |
 | --- | --- | --- | --- |
+| Lithos Metal source guide | Task compiler/fusion, GPU-resident generation/DSpark state and exact serving/cache ownership; candidate | [Skill](repositories/lithos-metal/SKILL.md) | [Record](../repo_packs/inferrsi/knowledge/common/skills/lithos-metal.json) |
 | ComfyUI source audit | Workflow-output cache validity and model-weight residency; source-derived candidate | [Skill](repositories/comfy-ui/SKILL.md) | [Record](../repo_packs/inferrsi/knowledge/common/skills/comfy-ui.json) |
 | stable-diffusion.cpp source audit | Native image weight/runtime capacity and protected segment prefetch; source-derived candidate | [Skill](repositories/stable-diffusion-cpp/audit-sd-cpp-residency/SKILL.md) | [Record](../repo_packs/inferrsi/knowledge/common/skills/audit-sd-cpp-residency.json) |
 | h3.c / MiniMax-H3 source audit | Conditioning, AdaLN precompute, phased/cached residency, precision and BF16 SSD streaming; source-derived candidate | [Skill](repositories/h3-c/audit-h3-runtime-residency/SKILL.md) | [Record](../repo_packs/inferrsi/knowledge/common/skills/audit-h3-runtime-residency.json) |
diff --git a/src/inferrsi/data/skill_library/repositories/lithos-metal/LICENSE b/src/inferrsi/data/skill_library/repositories/lithos-metal/LICENSE
new file mode 100644
index 000000000..d64569567
--- /dev/null
+++ b/src/inferrsi/data/skill_library/repositories/lithos-metal/LICENSE
@@ -0,0 +1,202 @@
+
+                                 Apache License
+                           Version 2.0, January 2004
+                        http://www.apache.org/licenses/
+
+   TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
+
+   1. Definitions.
+
+      "License" shall mean the terms and conditions for use, reproduction,
+      and distribution as defined by Sections 1 through 9 of this document.
+
+      "Licensor" shall mean the copyright owner or entity authorized by
+      the copyright owner that is granting the License.
+
+      "Legal Entity" shall mean the union of the acting entity and all
+      other entities that control, are controlled by, or are under common
+      control with that entity. For the purposes of this definition,
+      "control" means (i) the power, direct or indirect, to cause the
+      direction or management of such entity, whether by contract or
+      otherwise, or (ii) ownership of fifty percent (50%) or more of the
+      outstanding shares, or (iii) beneficial ownership of such entity.
+
+      "You" (or "Your") shall mean an individual or Legal Entity
+      exercising permissions granted by this License.
+
+      "Source" form shall mean the preferred form for making modifications,
+      including but not limited to software source code, documentation
+      source, and configuration files.
+
+      "Object" form shall mean any form resulting from mechanical
+      transformation or translation of a Source form, including but
+      not limited to compiled object code, generated documentation,
+      and conversions to other media types.
+
+      "Work" shall mean the work of authorship, whether in Source or
+      Object form, made available under the License, as indicated by a
+      copyright notice that is included in or attached to the work
+      (an example is provided in the Appendix below).
+
+      "Derivative Works" shall mean any work, whether in Source or Object
+      form, that is based on (or derived from) the Work and for which the
+      editorial revisions, annotations, elaborations, or other modifications
+      represent, as a whole, an original work of authorship. For the purposes
+      of this License, Derivative Works shall not include works that remain
+      separable from, or merely link (or bind by name) to the interfaces of,
+      the Work and Derivative Works thereof.
+
+      "Contribution" shall mean any work of authorship, including
+      the original version of the Work and any modifications or additions
+      to that Work or Derivative Works thereof, that is intentionally
+      submitted to Licensor for inclusion in the Work by the copyright owner
+      or by an individual or Legal Entity authorized to submit on behalf of
+      the copyright owner. For the purposes of this definition, "submitted"
+      means any form of electronic, verbal, or written communication sent
+      to the Licensor or its representatives, including but not limited to
+      communication on electronic mailing lists, source code control systems,
+      and issue tracking systems that are managed by, or on behalf of, the
+      Licensor for the purpose of discussing and improving the Work, but
+      excluding communication that is conspicuously marked or otherwise
+      designated in writing by the copyright owner as "Not a Contribution."
+
+      "Contributor" shall mean Licensor and any individual or Legal Entity
+      on behalf of whom a Contribution has been received by Licensor and
+      subsequently incorporated within the Work.
+
+   2. Grant of Copyright License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      copyright license to reproduce, prepare Derivative Works of,
+      publicly display, publicly perform, sublicense, and distribute the
+      Work and such Derivative Works in Source or Object form.
+
+   3. Grant of Patent License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      (except as stated in this section) patent license to make, have made,
+      use, offer to sell, sell, import, and otherwise transfer the Work,
+      where such license applies only to those patent claims licensable
+      by such Contributor that are necessarily infringed by their
+      Contribution(s) alone or by combination of their Contribution(s)
+      with the Work to which such Contribution(s) was submitted. If You
+      institute patent litigation against any entity (including a
+      cross-claim or counterclaim in a lawsuit) alleging that the Work
+      or a Contribution incorporated within the Work constitutes direct
+      or contributory patent infringement, then any patent licenses
+      granted to You under this License for that Work shall terminate
+      as of the date such litigation is filed.
+
+   4. Redistribution. You may reproduce and distribute copies of the
+      Work or Derivative Works thereof in any medium, with or without
+      modifications, and in Source or Object form, provided that You
+      meet the following conditions:
+
+      (a) You must give any other recipients of the Work or
+          Derivative Works a copy of this License; and
+
+      (b) You must cause any modified files to carry prominent notices
+          stating that You changed the files; and
+
+      (c) You must retain, in the Source form of any Derivative Works
+          that You distribute, all copyright, patent, trademark, and
+          attribution notices from the Source form of the Work,
+          excluding those notices that do not pertain to any part of
+          the Derivative Works; and
+
+      (d) If the Work includes a "NOTICE" text file as part of its
+          distribution, then any Derivative Works that You distribute must
+          include a readable copy of the attribution notices contained
+          within such NOTICE file, excluding those notices that do not
+          pertain to any part of the Derivative Works, in at least one
+          of the following places: within a NOTICE text file distributed
+          as part of the Derivative Works; within the Source form or
+          documentation, if provided along with the Derivative Works; or,
+          within a display generated by the Derivative Works, if and
+          wherever such third-party notices normally appear. The contents
+          of the NOTICE file are for informational purposes only and
+          do not modify the License. You may add Your own attribution
+          notices within Derivative Works that You distribute, alongside
+          or as an addendum to the NOTICE text from the Work, provided
+          that such additional attribution notices cannot be construed
+          as modifying the License.
+
+      You may add Your own copyright statement to Your modifications and
+      may provide additional or different license terms and conditions
+      for use, reproduction, or distribution of Your modifications, or
+      for any such Derivative Works as a whole, provided Your use,
+      reproduction, and distribution of the Work otherwise complies with
+      the conditions stated in this License.
+
+   5. Submission of Contributions. Unless You explicitly state otherwise,
+      any Contribution intentionally submitted for inclusion in the Work
+      by You to the Licensor shall be under the terms and conditions of
+      this License, without any additional terms or conditions.
+      Notwithstanding the above, nothing herein shall supersede or modify
+      the terms of any separate license agreement you may have executed
+      with Licensor regarding such Contributions.
+
+   6. Trademarks. This License does not grant permission to use the trade
+      names, trademarks, service marks, or product names of the Licensor,
+      except as required for reasonable and customary use in describing the
+      origin of the Work and reproducing the content of the NOTICE file.
+
+   7. Disclaimer of Warranty. Unless required by applicable law or
+      agreed to in writing, Licensor provides the Work (and each
+      Contributor provides its Contributions) on an "AS IS" BASIS,
+      WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
+      implied, including, without limitation, any warranties or conditions
+      of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
+      PARTICULAR PURPOSE. You are solely responsible for determining the
+      appropriateness of using or redistributing the Work and assume any
+      risks associated with Your exercise of permissions under this License.
+
+   8. Limitation of Liability. In no event and under no legal theory,
+      whether in tort (including negligence), contract, or otherwise,
+      unless required by applicable law (such as deliberate and grossly
+      negligent acts) or agreed to in writing, shall any Contributor be
+      liable to You for damages, including any direct, indirect, special,
+      incidental, or consequential damages of any character arising as a
+      result of this License or out of the use or inability to use the
+      Work (including but not limited to damages for loss of goodwill,
+      work stoppage, computer failure or malfunction, or any and all
+      other commercial damages or losses), even if such Contributor
+      has been advised of the possibility of such damages.
+
+   9. Accepting Warranty or Additional Liability. While redistributing
+      the Work or Derivative Works thereof, You may choose to offer,
+      and charge a fee for, acceptance of support, warranty, indemnity,
+      or other liability obligations and/or rights consistent with this
+      License. However, in accepting such obligations, You may act only
+      on Your own behalf and on Your sole responsibility, not on behalf
+      of any other Contributor, and only if You agree to indemnify,
+      defend, and hold each Contributor harmless for any liability
+      incurred by, or claims asserted against, such Contributor by reason
+      of your accepting any such warranty or additional liability.
+
+   END OF TERMS AND CONDITIONS
+
+   APPENDIX: How to apply the Apache License to your work.
+
+      To apply the Apache License to your work, attach the following
+      boilerplate notice, with the fields enclosed by brackets "[]"
+      replaced with your own identifying information. (Don't include
+      the brackets!)  The text should be enclosed in the appropriate
+      comment syntax for the file format. We also recommend that a
+      file or class name and description of purpose be included on the
+      same "printed page" as the copyright notice for easier
+      identification within third-party archives.
+
+   Copyright [yyyy] [name of copyright owner]
+
+   Licensed under the Apache License, Version 2.0 (the "License");
+   you may not use this file except in compliance with the License.
+   You may obtain a copy of the License at
+
+       http://www.apache.org/licenses/LICENSE-2.0
+
+   Unless required by applicable law or agreed to in writing, software
+   distributed under the License is distributed on an "AS IS" BASIS,
+   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+   See the License for the specific language governing permissions and
+   limitations under the License.
diff --git a/src/inferrsi/data/skill_library/repositories/lithos-metal/SKILL.md b/src/inferrsi/data/skill_library/repositories/lithos-metal/SKILL.md
new file mode 100644
index 000000000..ef7537608
--- /dev/null
+++ b/src/inferrsi/data/skill_library/repositories/lithos-metal/SKILL.md
@@ -0,0 +1,93 @@
+---
+name: lithos-metal
+description: "Inspect Lithos Metal compilation and fusion, GPU-resident generation and DSpark state, or local serving and exact-prefix caches on Apple silicon; use when changing recipes, diagnosing stale state, or qualifying an inference claim."
+---
+
+# Lithos Metal operating guide
+
+## Trigger and boundary
+
+Use this entry for Lithos Metal compiler, generation-state and serving/cache
+questions. Route the concrete question before suggesting a recipe or reset.
+This is a **source-derived candidate** at upstream commit
+`6653b7df5f2c4392ec7b038960f2796b1499923f`, distribution `lithos-metal` 0.1.2,
+Python import namespace `monolith`. Empirical efficacy validation is **not_run**.
+
+Source facts describe that snapshot; the decision procedures are proposed
+applications of those facts. Package/CPU contracts and a synthetic Metal program
+can be checked without model downloads. They do not qualify complete model
+continuation, DSpark numerics, model quality or service performance.
+
+## Choose a route
+
+| Question | Read next | Output |
+| --- | --- | --- |
+| Missing op binding, fusion eligibility, tile/workers, barriers or scratch aliasing | [compile-fusion](sub-skills/compile-fusion/SKILL.md) | Region/dependency ledger, normalized control and fallback checks |
+| Wrong continuation after rejection, active rows, ring overflow, stop or cancellation | [generation-state](sub-skills/generation-state/SKILL.md) | Committed-prefix/state ledger and bounded reproduction plan |
+| Local startup, packs, HTTP state, prefix hits, disk integrity or idle reload | [serving-cache](sub-skills/serving-cache/SKILL.md) | Cache-identity/lifecycle decision and isolated checks |
+
+A session replacement can cross all three routes: a recipe changes program
+geometry, generation owns committed state, and serving owns adoption/cache
+identity. Inspect that boundary before interpreting a cache hit as correctness.
+
+## Establish the package contract
+
+1. Record the actual source revision, package version, backend ID/configuration,
+   chip identity and core count, OS, target/draft checkpoint identities, context,
+   prompt token IDs, sampling and numerical options. Redact private paths and
+   prompt content when sharing. Keep absent fields explicit.
+2. Read [repo-provenance](references/repo-provenance.md) and the pinned
+   [source readset](references/source-notes.md). Revision or API drift requires
+   a fresh bounded inspection; do not transfer recipe qualification by name.
+3. Read [package and API contract](references/package-api.md) for installation,
+   namespaces, optional dependencies and CPU-versus-Metal boundaries.
+4. Optionally run the bundled [inspect_contract.py](scripts/inspect_contract.py)
+   in the selected installed environment. It reports software/API facts;
+   `--check-contract` adds tiny CPU checks. It never creates a device, downloads
+   weights, starts a server or runs automatically when this skill is read.
+
+```bash
+python <SKILL_DIR>/scripts/inspect_contract.py --check-contract
+lithos-metal --version
+lithos-metal serve --help
+```
+
+`<SKILL_DIR>` is this packaged skill directory. Locate the resource through the
+library reader or package resources; the helper works from any current directory.
+A successful diagnostic is limited to its reported checks.
+
+## Work sequence
+
+1. Name the affected owner: model graph, chip recipe, emitted `Program`, runtime
+   buffers/control, serving session, pack cache or prefix checkpoint.
+2. Reconstruct the earliest boundary whose inputs/outputs disagree. Use the
+   relevant route's counter-signals before proposing a mutation.
+3. Preserve an unfused/native or target-only control where applicable, matching
+   decoded weights, numerical mode, row bounds and sampling policy.
+4. Verify control/dependency and persistent-state correctness before timing.
+   Separate source inspection, CPU fixtures, synthetic Metal execution,
+   model continuation and HTTP workload evidence in the result.
+5. Report the smallest supported finding, missing checks and next bounded action.
+   Use campaign-owned frozen gates for any efficacy claim. Missing required
+   model or E2E evidence remains unverified/`INCONCLUSIVE` for that claim.
+
+## Counter-signals and limits
+
+Layer-wise mixer fusion is not whole-model fusion; MLP regions can remain
+separate. Fewer dispatches do not establish a faster complete round. A measured
+recipe for one chip/core variant is not another variant's default. CPU cache
+tests do not demonstrate GPU state rollback. `/health` liveness does not show a
+model is loaded, correct or warm. A target-compute projection is not an E2E rate.
+
+Read [troubleshooting](references/troubleshooting.md) for package/import and
+cross-route failures. Source code and documentation occasionally differ in
+generality: use the precise code path and preserve the documentation's scope.
+No model downloads, device resets, live cache deletion, server launch or client
+execution are implied by reading this guide.
+
+## Attribution and reuse
+
+Read [notices](references/notices.md) when adapting material. The bundled
+[LICENSE](LICENSE), [upstream NOTICE](references/upstream-notice.md) and
+[MLX license](references/mlx-license.md) preserve source attribution. This guide
+adds no runtime dependency on Lithos Metal, MPK/Mirage, MLX or DSpark references.
diff --git a/src/inferrsi/data/skill_library/repositories/lithos-metal/metadata.json b/src/inferrsi/data/skill_library/repositories/lithos-metal/metadata.json
new file mode 100644
index 000000000..6136acb38
--- /dev/null
+++ b/src/inferrsi/data/skill_library/repositories/lithos-metal/metadata.json
@@ -0,0 +1,64 @@
+{
+  "profile": "local-skill-v1",
+  "schema_version": 1,
+  "kind": "skill",
+  "origin": [
+    "external"
+  ],
+  "status": "candidate",
+  "empirical_validation": "not_run",
+  "tags": [
+    "lithos-metal",
+    "metal",
+    "apple-silicon",
+    "repository-guide"
+  ],
+  "sources": [
+    {
+      "type": "git",
+      "url": "https://github.com/lithos-ai/lithos-metal",
+      "commit": "6653b7df5f2c4392ec7b038960f2796b1499923f",
+      "paths": [
+        "kernels/common/spec_ops.metal",
+        "monolith/backends/metal/base.py",
+        "monolith/backends/metal/registry.py",
+        "monolith/compiler/arena.py",
+        "monolith/compiler/barriers.py",
+        "monolith/compiler/coverage.py",
+        "monolith/compiler/decoder_fusion.py",
+        "monolith/compiler/emit.py",
+        "monolith/compiler/region_fusion.py",
+        "monolith/compiler/static_fusion.py",
+        "monolith/core/step_state.py",
+        "monolith/generate.py",
+        "monolith/runtime/engine.py",
+        "monolith/runtime/prefix_cache.py",
+        "monolith/runtime/prefix_store.py",
+        "monolith/serve.py",
+        "monolith/serving/cache.py",
+        "monolith/serving/lifecycle.py",
+        "monolith/serving/setup.py",
+        "monolith/spec/dspark/model.py",
+        "runtime/src/metal_core.mm"
+      ],
+      "readset": "references/source-notes.md"
+    }
+  ],
+  "applicability": {
+    "scope": "Pinned Lithos Metal 0.1.2 source/API contracts; Apple silicon Metal runtime; model and recipe qualification required separately."
+  },
+  "limitations": [
+    "Source-derived candidate; empirical efficacy validation not_run.",
+    "CPU/software and synthetic Metal contracts do not qualify full-model/DSpark numerics, quality or E2E performance.",
+    "No model downloads, live DisCo deployment or inference backend integration is provided."
+  ],
+  "provenance_readsets": [
+    "references/source-notes.md"
+  ],
+  "consumption_mode": "packaged",
+  "license": "Apache-2.0",
+  "disable-model-invocation": true,
+  "metadata": {
+    "disco-role": "operating"
+  }
+}
diff --git a/src/inferrsi/data/skill_library/repositories/lithos-metal/references/mlx-license.md b/src/inferrsi/data/skill_library/repositories/lithos-metal/references/mlx-license.md
new file mode 100644
index 000000000..98ff47b9e
--- /dev/null
+++ b/src/inferrsi/data/skill_library/repositories/lithos-metal/references/mlx-license.md
@@ -0,0 +1,21 @@
+MIT License
+
+Copyright © 2023 Apple Inc.
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.
diff --git a/src/inferrsi/data/skill_library/repositories/lithos-metal/references/notices.md b/src/inferrsi/data/skill_library/repositories/lithos-metal/references/notices.md
new file mode 100644
index 000000000..c29013d49
--- /dev/null
+++ b/src/inferrsi/data/skill_library/repositories/lithos-metal/references/notices.md
@@ -0,0 +1,22 @@
+# Source notices
+
+This candidate distills `lithos-ai/lithos-metal` at
+`6653b7df5f2c4392ec7b038960f2796b1499923f` (Apache-2.0). The owner-local
+[LICENSE](../LICENSE) is preserved from that snapshot. The
+[upstream third-party registry](upstream-notice.md) and [MLX license](mlx-license.md)
+are preserved verbatim; retaining them does not claim that this skill copies
+every listed implementation. No Metal kernel or model weights are bundled.
+
+The task compiler/runtime implementation is standalone. Upstream acknowledges
+MPK/Mirage design influence, attributed registry/module/model fragments,
+MLX-derived format/kernel fragments, Transformers numerical conventions and
+DeepSpec DSpark semantics. Those provenance relationships must not be erased
+when adapting upstream code. This guide's small software diagnostic is new
+glue around the pinned public Python contracts, not an upstream script rename.
+
+The extraction/verification procedure was informed by VectorSpaceLab/AREX-Skill
+at `ac3fe1afa80fb9a09775ecfb2b6cc3ba850a2db6`, specifically its
+`create-repo-skill`, `prepare-repo-skill-env` and `verify-repo-skill` workflows.
+The user's contribution destination is InferRSI, so this graph uses InferRSI's
+registry and two matrices. It makes no live DisCo import, router deployment or
+cross-agent export claim.
diff --git a/src/inferrsi/data/skill_library/repositories/lithos-metal/references/package-api.md b/src/inferrsi/data/skill_library/repositories/lithos-metal/references/package-api.md
new file mode 100644
index 000000000..f8e2d9bfb
--- /dev/null
+++ b/src/inferrsi/data/skill_library/repositories/lithos-metal/references/package-api.md
@@ -0,0 +1,70 @@
+# Package and API contract
+
+Use for installation planning and for distinguishing software inspection from
+accelerator/model verification. Facts below are from the pinned source readset;
+selected API signatures and CLI help were also inspected in an installed 0.1.2.
+
+## Installation and dependencies
+
+The distribution and public command are `lithos-metal`; imports use `monolith`.
+`pyproject.toml` declares Python >=3.10 and NumPy >=1.26. The source build/release
+instructions use Python 3.12 and Xcode Command Line Tools. These are distinct
+claims: metadata acceptance of a Python version is not a tested native build.
+The source build uses CMake and nanobind on Darwin and packages the Metal kernel
+sources. A non-Darwin package install omits the native extension.
+
+Select an already provisioned environment or a private environment suited to the
+requested scope. An explicit install, when authorized, can use:
+
+```bash
+python -m pip install 'lithos-metal==0.1.2'
+python -c 'import monolith; print(monolith.__version__)'
+```
+
+Installing a package can access the package index and build native code. It is
+separate from running the no-download diagnostic. Availability of a compatible
+prebuilt distribution must be checked in the chosen environment. The upstream
+alternative is the public Homebrew tap; it is not needed to read this skill.
+
+| Scope | Declared dependency surface | What it can establish |
+| --- | --- | --- |
+| Core package | NumPy | Imports, IR/layout/program/pack/cache Python contracts |
+| `dev` extra | pytest | Source contract tests with tiny fixtures |
+| `serve` extra | FastAPI, Pydantic, Uvicorn, Transformers, safetensors, huggingface_hub, Jinja2 | HTTP adapter and model/tokenizer loading surfaces; model startup still requires assets and Metal |
+| `oracle` extra | Torch, Transformers, safetensors, huggingface_hub | Reference comparisons, not automatic golden/model availability |
+
+FastAPI TestClient tests additionally need `httpx`; it is not a listed `serve`
+dependency at this pin. Do not install all extras merely to inspect the package.
+Apple silicon/macOS 26+ and a matching registered backend are the documented
+runtime scope. Native module import and a tiny shader are separate checks;
+neither proves a specific target/draft recipe runs correctly.
+
+## Public call boundaries
+
+| Entry | Relevant parameters and defaults at 0.1.2 | Contract |
+| --- | --- | --- |
+| `monolith.generate.load_session(model_dir, pack_dir, *, max_context=4096, drafter_dir=None, drafter_pack=None, drafter_kind='dspark', **options)` | Local checkpoint and prepared pack inputs; target/draft config resolved by model packages | Constructs `Session`; it is not the serving checkpoint downloader |
+| `monolith.generate.Session(...)` | `verify='cost'`, `verify_length=None`, `draft_sampling='argmax'`, `barriers='minimal'`, `commute_norm=True`, `fast_math=False`, `prefix_cache=False` | Lower-level policy; serving overrides several choices |
+| `Session.generate(prompt_ids, max_new_tokens, *, steps_per_cb=8, in_flight=3, on_tokens=None, cancelled=None, cache_prefix_tokens=0)` | Nonempty integer prompt IDs; callback receives cumulative tokens | Handles prefill, cache restore, decoder transition and bounded host pumps |
+| `monolith.compiler.compile_program(...)` | `dynamic_t=False`, `drafter=None`, `barriers='minimal'`, `commute_norm=True` | Speculation requires dynamic T, compatible layout bounds and a draft pack |
+| `monolith.runtime.Engine(program, device=None, buffers=None, fast_math=False, pipeline_cache=None)` | Allocates/maps buffers and encodes ICB; constructor creates a device if absent | Requires actual native runtime; inspection need not construct it |
+| `StepStateLayout(t_max=8, gamma_max=7)` | Requires `t_max >= gamma_max + 1 >= 1` | Shared binary layout, not a model capacity or recipe guarantee |
+| `PrefixCache(entries, max_bytes=4*1024**3, min_tokens=0, store=None)` | Host snapshots, optional disk store | Saves declared model/drafter state; scratch and worker counters are excluded |
+
+Do not copy the illustrative defaults into campaign thresholds. `Session`'s
+cost policy is not the server default: `ServingAssets.options` selects fixed
+verification for its DSpark path. Read the relevant subskill before choosing.
+
+## Safe software inspection
+
+The bundled helper accepts `--expected-version VERSION` (default `0.1.2`) and
+`--check-contract`. Output is JSON with version agreement, API parameter names,
+optional-module discovery and, if requested, CPU layout/barrier assertions.
+Exit 0 means those checks completed; exit 1 means a required import, version or
+contract check failed. It omits installation paths and exception text. It does
+not report overall backend readiness and never substitutes for Metal/model tests.
+
+CLI `--version` and `serve --help` are parser checks. `serve --model HUB_ID` may
+download target and automatic draft checkpoints, prepare large packs, compile
+and warm up. Use the local-only recipe in the serving route only when a server
+run is separately within scope.
diff --git a/src/inferrsi/data/skill_library/repositories/lithos-metal/references/repo-provenance.md b/src/inferrsi/data/skill_library/repositories/lithos-metal/references/repo-provenance.md
new file mode 100644
index 000000000..557719841
--- /dev/null
+++ b/src/inferrsi/data/skill_library/repositories/lithos-metal/references/repo-provenance.md
@@ -0,0 +1,67 @@
+# Repository provenance
+
+Use this snapshot to detect drift before applying the guide. Source citations
+are research provenance, not campaign baseline/candidate defaults.
+
+```json
+{
+  "schema": "disco.repo-provenance.v1",
+  "generated_at_utc": "2026-10-23T09:09:46Z",
+  "repository": {
+    "name": "lithos-metal",
+    "remote_url": "https://github.com/lithos-ai/lithos-metal",
+    "vcs": "git",
+    "branch": "main",
+    "tag": null,
+    "commit": "6653b7df5f2c4392ec7b038960f2796b1499923f",
+    "working_tree": "clean",
+    "dirty_paths": []
+  },
+  "packages": [{"name": "lithos-metal", "version": "0.1.2", "import_names": ["monolith"]}],
+  "evidence": {
+    "source_roots": ["monolith/compiler", "monolith/core", "monolith/runtime", "monolith/spec", "monolith/serving", "runtime/src", "kernels/common"],
+    "docs": ["README.md", "CONTRIBUTING.md", "docs/design"],
+    "examples": [],
+    "tests": ["tests/contract", "tests/runtime/test_toy_program.py"],
+    "configs": ["pyproject.toml", "setup.py", "monolith/backends/metal/registry.py"]
+  }
+}
+```
+
+The [bounded source readset](source-notes.md) specifies inspected sections and
+symbols rather than claiming complete directory coverage. Selected installed
+API and CLI facts were inspected at 0.1.2; the source Python metadata minimum
+and source-build interpreter guidance remain distinct. Candidate efficacy is
+`not_run`; source inspection, CPU contracts and synthetic Metal checks do not
+promote this guide or establish full-model/DSpark/service qualification.
+
+Refresh when the upstream revision/dirty state, package version, public API,
+backend recipe, layout or source file ownership changes. Compare changed
+contracts and regenerate dependent guidance; do not merely replace the pin.
+Missing source/runtime evidence stays unavailable rather than resolving to a
+newer live implementation silently.
+
+## Extraction boundary
+
+Included: compiler coverage/regions/barriers/arena, Program and StepState ABI,
+generation orchestration/DSpark commit, backend identity, serving assets,
+pack/prefix validity and idle/session lifetime. Source tests are evidence and
+separately selectable validation candidates; no tests are run by skill reads.
+
+Excluded: model catalog completeness, new model/format implementations, MoE
+qualification, exhaustive kernel tuning, external oracle code/weights, CI/release
+automation, benchmark archives and client launchers. The owner has no complete
+Apple GPU support or service-readiness claim. Exact citations to model-related
+contracts do not imply those models were loaded or compared.
+
+InferRSI `local-skill-v1` keeps only name/description in SKILL frontmatter.
+The AREX license, `disable-model-invocation` and `metadata.disco-role` fields
+are preserved in each sidecar and exposed by the effective metadata reader.
+This destination normalization replaces AREX's raw-frontmatter gate; it does
+not claim direct DisCo import compatibility.
+
+AREX's repo2skill structure, source/script maps, installed inspection and
+assertion-backed verification were adapted for an InferRSI candidate. InferRSI
+owns discovery via its registry/catalog and two matrices. No DisCo fixed-taxonomy
+projection or live importer is needed for this destination; review-only scope,
+native candidates and usability artifacts stay outside the runtime skill tree.
diff --git a/src/inferrsi/data/skill_library/repositories/lithos-metal/references/source-notes.md b/src/inferrsi/data/skill_library/repositories/lithos-metal/references/source-notes.md
new file mode 100644
index 000000000..a49fdb127
--- /dev/null
+++ b/src/inferrsi/data/skill_library/repositories/lithos-metal/references/source-notes.md
@@ -0,0 +1,87 @@
+# Pinned source readset
+
+All Lithos links below resolve to
+`6653b7df5f2c4392ec7b038960f2796b1499923f`. **Code** means inspected
+implementation/contract; **Docs** means stated intent, support or measurement
+scope; **Tests** means inspected assertions, not a claim that every test ran.
+The procedures are proposed uses of this evidence. No upstream metric is a
+local measurement, and no source test implies full-model qualification.
+
+## Shared package and provenance
+
+| Source and inspected scope | Evidence retained |
+| --- | --- |
+| [pyproject.toml L1–36](https://github.com/lithos-ai/lithos-metal/blob/6653b7df5f2c4392ec7b038960f2796b1499923f/pyproject.toml#L1-L36); [setup.py L1–36](https://github.com/lithos-ai/lithos-metal/blob/6653b7df5f2c4392ec7b038960f2796b1499923f/setup.py#L1-L36) | Code: distribution/version/Python requirement, extras, entry point, Darwin extension and bundled kernels |
+| [monolith/cli.py L1–44](https://github.com/lithos-ai/lithos-metal/blob/6653b7df5f2c4392ec7b038960f2796b1499923f/monolith/cli.py#L1-L44); [runtime/__init__.py L1–23](https://github.com/lithos-ai/lithos-metal/blob/6653b7df5f2c4392ec7b038960f2796b1499923f/monolith/runtime/__init__.py#L1-L23) | Code: CLI dispatch, namespace, optional native import; import availability differs from execution |
+| [README.md](https://github.com/lithos-ai/lithos-metal/blob/6653b7df5f2c4392ec7b038960f2796b1499923f/README.md); [CONTRIBUTING.md L1–40](https://github.com/lithos-ai/lithos-metal/blob/6653b7df5f2c4392ec7b038960f2796b1499923f/CONTRIBUTING.md#L1-L40) | Docs: startup/download behavior, model/hardware status, source-build Python 3.12 and Command Line Tools; performance section's target-compute scope |
+| [LICENSE](https://github.com/lithos-ai/lithos-metal/blob/6653b7df5f2c4392ec7b038960f2796b1499923f/LICENSE); [third_party/NOTICE](https://github.com/lithos-ai/lithos-metal/blob/6653b7df5f2c4392ec7b038960f2796b1499923f/third_party/NOTICE); [MLX-LICENSE](https://github.com/lithos-ai/lithos-metal/blob/6653b7df5f2c4392ec7b038960f2796b1499923f/third_party/MLX-LICENSE) | Attribution retained verbatim in owner-local resources |
+
+## Compiler/fusion route
+
+| Source, symbol and inspected lines | Evidence retained |
+| --- | --- |
+| [coverage.py L1–32](https://github.com/lithos-ai/lithos-metal/blob/6653b7df5f2c4392ec7b038960f2796b1499923f/monolith/compiler/coverage.py#L1-L32), `check_coverage` | Code: missing binding fails before emission |
+| [emit.py L1390–1480](https://github.com/lithos-ai/lithos-metal/blob/6653b7df5f2c4392ec7b038960f2796b1499923f/monolith/compiler/emit.py#L1390-L1480), `emit_program`; [L1534–1620](https://github.com/lithos-ai/lithos-metal/blob/6653b7df5f2c4392ec7b038960f2796b1499923f/monolith/compiler/emit.py#L1534-L1620), `lower_round`, `compile_program` | Code: roles/bounds, tail, backend finalize, dynamic speculative preconditions, commit ordering, policy fallback |
+| [region_fusion.py](https://github.com/lithos-ai/lithos-metal/blob/6653b7df5f2c4392ec7b038960f2796b1499923f/monolith/compiler/region_fusion.py), `fuse_regions`, `merge_labeled_regions` | Code: disjoint regions, normalized control, merge selection, StepState timeout guard |
+| [static_fusion.py L53–210](https://github.com/lithos-ai/lithos-metal/blob/6653b7df5f2c4392ec7b038960f2796b1499923f/monolith/compiler/static_fusion.py#L53-L210), `normalize`; [L929–1185](https://github.com/lithos-ai/lithos-metal/blob/6653b7df5f2c4392ec7b038960f2796b1499923f/monolith/compiler/static_fusion.py#L929-L1185), `merge` | Code: option/geometry guards and fixed-worker/task/barrier source construction; middle normalization transforms not exhaustively audited |
+| [decoder_fusion.py L1–68](https://github.com/lithos-ai/lithos-metal/blob/6653b7df5f2c4392ec7b038960f2796b1499923f/monolith/compiler/decoder_fusion.py#L1-L68), `optimize`, `elide_dead_gdn_state` | Code: mixer/MLP boundaries and commit-rewritten state-store ownership |
+| [barriers.py L1–74](https://github.com/lithos-ai/lithos-metal/blob/6653b7df5f2c4392ec7b038960f2796b1499923f/monolith/compiler/barriers.py#L1-L74), `place_barriers`; [arena.py L1–62](https://github.com/lithos-ai/lithos-metal/blob/6653b7df5f2c4392ec7b038960f2796b1499923f/monolith/compiler/arena.py#L1-L62), `reuse_arenas` | Code: conservative whole-buffer hazards, variants, cross-step boundary and physical scratch lifetime exclusions |
+| [passes/fuse_norm.py L1–33](https://github.com/lithos-ai/lithos-metal/blob/6653b7df5f2c4392ec7b038960f2796b1499923f/monolith/compiler/passes/fuse_norm.py#L1-L33), `fuse_norm_stat` | Code: BF16 producer statistic hoisting; distinct from commuted norm |
+| [metal/base.py L1–85](https://github.com/lithos-ai/lithos-metal/blob/6653b7df5f2c4392ec7b038960f2796b1499923f/monolith/backends/metal/base.py#L1-L85); [registry.py L1–61](https://github.com/lithos-ai/lithos-metal/blob/6653b7df5f2c4392ec7b038960f2796b1499923f/monolith/backends/metal/registry.py#L1-L61) | Code: backend/device identity, cache digest, conservative base hooks, chip/family/core selection; no exhaustive recipe qualification |
+| [design.md](https://github.com/lithos-ai/lithos-metal/blob/6653b7df5f2c4392ec7b038960f2796b1499923f/docs/design/design.md), compilation/fusion/storage/numerics sections; [mixers.md](https://github.com/lithos-ai/lithos-metal/blob/6653b7df5f2c4392ec7b038960f2796b1499923f/docs/design/mixers.md), input-normalization section | Docs: layer-wise scope, separate MLP, worker semantics, intentional commuted-norm rounding |
+
+## Generation route
+
+| Source, symbol and inspected lines | Evidence retained |
+| --- | --- |
+| [step_state.py L1–119](https://github.com/lithos-ai/lithos-metal/blob/6653b7df5f2c4392ec7b038960f2796b1499923f/monolith/core/step_state.py#L1-L119), `StepStateLayout` | Code: complete ABI fields, bounds, alignment, host serialization |
+| [program.py](https://github.com/lithos-ai/lithos-metal/blob/6653b7df5f2c4392ec7b038960f2796b1499923f/monolith/runtime/program.py), spec dataclasses and `Program`; [engine.py L31–122](https://github.com/lithos-ai/lithos-metal/blob/6653b7df5f2c4392ec7b038960f2796b1499923f/monolith/runtime/engine.py#L31-L122), `compile_pipelines`, `Engine` | Code: buffer/pipeline identity, program-specific params, ICB/run/profile boundaries and timeout reporting |
+| [generate.py L81–112](https://github.com/lithos-ai/lithos-metal/blob/6653b7df5f2c4392ec7b038960f2796b1499923f/monolith/generate.py#L81-L112), `Session.__init__`; [L475–697](https://github.com/lithos-ai/lithos-metal/blob/6653b7df5f2c4392ec7b038960f2796b1499923f/monolith/generate.py#L475-L697), `reset`, `generate`, `_accept_stats`; [L741–774](https://github.com/lithos-ai/lithos-metal/blob/6653b7df5f2c4392ec7b038960f2796b1499923f/monolith/generate.py#L741-L774), `load_session` | Code: prefill/cached tails, exact reference fallback, logical KV validity, context checks, pump/callback/cancel and productive rounds |
+| [spec_ops.metal L108–195](https://github.com/lithos-ai/lithos-metal/blob/6653b7df5f2c4392ec7b038960f2796b1499923f/kernels/common/spec_ops.metal#L108-L195), `verify_select`, `accept_scan`; [advance.metal L1–37](https://github.com/lithos-ai/lithos-metal/blob/6653b7df5f2c4392ec7b038960f2796b1499923f/kernels/common/advance.metal#L1-L37) | Code: anchor/row bounds, committed output, prompt branch, ring/context/stop bookkeeping; stochastic sampler itself not exhaustively audited |
+| [metal_core.mm L310–387](https://github.com/lithos-ai/lithos-metal/blob/6653b7df5f2c4392ec7b038960f2796b1499923f/runtime/src/metal_core.mm#L310-L387), `drain_ring`, `Runner::run`, `drain` | Code: sequence-tagged ring, bounded submissions and completion ownership |
+| [dspark/model.py L132–183](https://github.com/lithos-ai/lithos-metal/blob/6653b7df5f2c4392ec7b038960f2796b1499923f/monolith/spec/dspark/model.py#L132-L183), drafter construction; [L364–441](https://github.com/lithos-ai/lithos-metal/blob/6653b7df5f2c4392ec7b038960f2796b1499923f/monolith/spec/dspark/model.py#L364-L441), lowering/select/context update | Code: checkpoint/head ownership, target feature injection and no-op post-accept context hook; no external DeepSpec oracle inspection |
+| [speculative-decoding.md](https://github.com/lithos-ai/lithos-metal/blob/6653b7df5f2c4392ec7b038960f2796b1499923f/docs/design/speculative-decoding.md) | Docs: DSpark components, sampling distribution intent, fixed serving versus lower-level policies and target-only timing omissions |
+
+## Serving/cache route
+
+| Source, symbol and inspected lines | Evidence retained |
+| --- | --- |
+| [serving/cache.py L19–157](https://github.com/lithos-ai/lithos-metal/blob/6653b7df5f2c4392ec7b038960f2796b1499923f/monolith/serving/cache.py#L19-L157), resolve/identity/validate/ensure pack | Code: local-only resolution, shard metadata identity, structural checks, locks and atomic publication |
+| [serving/setup.py L22–142](https://github.com/lithos-ai/lithos-metal/blob/6653b7df5f2c4392ec7b038960f2796b1499923f/monolith/serving/setup.py#L22-L142), `ServingAssets.options`, `prepare` | Code: fixed verification, pair/shape/tap checks, internal draft capacity, recipes and precision |
+| [prefix_cache.py L1–169](https://github.com/lithos-ai/lithos-metal/blob/6653b7df5f2c4392ec7b038960f2796b1499923f/monolith/runtime/prefix_cache.py#L1-L169), `PrefixCache` | Code: strict exact matching, declared state only, pre-copy eviction, fit-before-write, stale drop and flush outcome |
+| [prefix_store.py L1–125](https://github.com/lithos-ai/lithos-metal/blob/6653b7df5f2c4392ec7b038960f2796b1499923f/monolith/runtime/prefix_store.py#L1-L125); [L136–320](https://github.com/lithos-ai/lithos-metal/blob/6653b7df5f2c4392ec7b038960f2796b1499923f/monolith/runtime/prefix_store.py#L136-L320); [L351–451](https://github.com/lithos-ai/lithos-metal/blob/6653b7df5f2c4392ec7b038960f2796b1499923f/monolith/runtime/prefix_store.py#L351-L451), store/read/write/header/blob/quota paths | Code: ownership lock, private files, per-blob integrity, queue, fsync/rename, pin-aware eviction and miss fallback |
+| [generate.py L285–378](https://github.com/lithos-ai/lithos-metal/blob/6653b7df5f2c4392ec7b038960f2796b1499923f/monolith/generate.py#L285-L378), `prefix_identity`, release/adopt | Code: code/device/config/tokenizer-extra identity, payload size/mtime limit, allocation compatibility and state reset |
+| [serve.py L103–263](https://github.com/lithos-ai/lithos-metal/blob/6653b7df5f2c4392ec7b038960f2796b1499923f/monolith/serve.py#L103-L263), select/store/close/load/unload; [L425–710](https://github.com/lithos-ai/lithos-metal/blob/6653b7df5f2c4392ec7b038960f2796b1499923f/monolith/serve.py#L425-L710), app/admission/stream/adapters | Code: one active generation, shared sequence cache, bounded flush and protocol/lifecycle surfaces; converter internals not exhaustively audited |
+| [lifecycle.py L129–263](https://github.com/lithos-ai/lithos-metal/blob/6653b7df5f2c4392ec7b038960f2796b1499923f/monolith/serving/lifecycle.py#L129-L263), load/release/probe/watch | Code: pending arrivals, GPU-lock recheck and failed-load sharing |
+| [serving.md](https://github.com/lithos-ai/lithos-metal/blob/6653b7df5f2c4392ec7b038960f2796b1499923f/docs/design/serving.md), startup/session/request/prefill/idle/prefix sections | Docs: operational intent and protocol limits; timing anecdotes are not imported measurements |
+
+## Inspected test assertions and script selection
+
+Tests inspected for the stated contracts: `tests/contract/test_barriers.py`,
+`test_arena.py`, `test_step_state.py`, `test_generate_speculative.py`,
+`test_prefix_cache.py`, `test_prefix_store.py`, `test_serve.py`,
+`test_serving_cache.py`, `test_idle_release.py`, and
+`tests/runtime/test_toy_program.py`. They provide tiny/mock cases for hazards,
+ABI, output limits/cancel, exact checkpoints, disk integrity and idle/HTTP
+ownership; the runtime file provides synthetic ICB/reencode/ring/pipeline checks.
+Their source presence is not an execution result or a full-model oracle.
+
+Source script inventory was bounded by public entry points and relevant tools.
+`monolith/cli.py`/`serve.py` parser and local-only behavior are distilled in the
+bundled package/workflow references. `tools/pack_weights.py` writes large derived
+packs; it is reference-only, with pack validity distilled here rather than a
+checkout-dependent wrapper. `tools/profile_writer.py` performs large synthetic
+GPU measurements and can write backend configurations; it is excluded from
+automatic diagnostics. `tools/bench/profile_spec_round.py` and
+`target_verify_latency.py` need model packs/accelerator execution and measure
+different intervals; their scope informs measurement cautions, but they are
+not bundled or recommended as automatic checks. Remaining benchmark, CI,
+release, golden-generation and export tools are outside the selected scope.
+The bundled `scripts/inspect_contract.py` adapts only safe public contract
+inspection; no original source script is required at runtime.
+
+AREX workflow provenance:
+[create-repo-skill](https://github.com/VectorSpaceLab/AREX-Skill/blob/ac3fe1afa80fb9a09775ecfb2b6cc3ba850a2db6/cli/packages/coding-agent/src/disco/skills/create-repo-skill/SKILL.md),
+[verify-repo-skill](https://github.com/VectorSpaceLab/AREX-Skill/blob/ac3fe1afa80fb9a09775ecfb2b6cc3ba850a2db6/cli/packages/coding-agent/src/disco/skills/verify-repo-skill/SKILL.md).
+Only the source/extraction/verification process is reused; no live DisCo
+deployment or fixed-taxonomy classification is claimed.
diff --git a/src/inferrsi/data/skill_library/repositories/lithos-metal/references/troubleshooting.md b/src/inferrsi/data/skill_library/repositories/lithos-metal/references/troubleshooting.md
new file mode 100644
index 000000000..0a74b2ee6
--- /dev/null
+++ b/src/inferrsi/data/skill_library/repositories/lithos-metal/references/troubleshooting.md
@@ -0,0 +1,19 @@
+# Cross-route troubleshooting
+
+Use before diagnosing a kernel or modifying stored state.
+
+| Symptom | Discriminating check | Recovery or stop condition |
+| --- | --- | --- |
+| `No module named lithos_metal` | Distribution is `lithos-metal`; namespace is `monolith` | Use the documented imports. Run the bundled software diagnostic in the selected installed environment. |
+| Import works, generation does not | Separate native-extension import, device creation, shader compilation, registered chip/config match and model assets | Inspect the actual failed boundary. A CPU-only install is useful for contract inspection but is not a generation backend. |
+| Source build fails | Compare interpreter/toolchain to the source Python 3.12/Command Line Tools instructions and Darwin build dependencies | Preserve the build error. Missing `xcrun metal` alone does not prove runtime shader compilation unavailable; test the actual native boundary separately when authorized. |
+| `serve` import or TestClient fails | `serve` extra versus separately required `httpx` in tests | Install only the scoped missing dependencies in an appropriate environment. No model weights are required for mocked HTTP tests. |
+| Package/version differs from this guide | Helper version/API report and pinned source readset | Treat as drift; inspect changed signatures and source before using old recipe advice. |
+| `No registered backend` or backend/device mismatch | Chip name, Apple family and core count all participate | Do not rename/copy another chip's profile. Use a real matching backend or stop qualification. |
+| Apparent huge throughput gain | Identify measured interval and counted committed tokens | Target-only compute excludes drafting, acceptance/rewind, prefill and HTTP overhead. Require complete workload evidence for service wording. |
+| State wrong only after session/recipe change | Route compiler geometry, generation commit and serving adoption together | Compare compatible state names/size/layout, identical weight windows and exact-prefix identity. Recompilation or a pipeline hit alone does not reset sequence state. |
+
+Keep reproducible observations separate from hypotheses. A source contract
+assertion supports that contract at the pin; it cannot attribute an unseen
+production failure or certify all backends. Reports should identify source,
+software/CPU, synthetic Metal and model/E2E coverage individually.
diff --git a/src/inferrsi/data/skill_library/repositories/lithos-metal/references/upstream-notice.md b/src/inferrsi/data/skill_library/repositories/lithos-metal/references/upstream-notice.md
new file mode 100644
index 000000000..a5f46f827
--- /dev/null
+++ b/src/inferrsi/data/skill_library/repositories/lithos-metal/references/upstream-notice.md
@@ -0,0 +1,94 @@
+# Third-party provenance
+
+Monolith is a standalone codebase (design D15). Code copied or adapted from other projects keeps its original license
+header and carries a one-line provenance comment naming the source repository, path and commit. This file is the
+registry of every such file or fragment. Add an entry in the same change that copies the code.
+
+Format: `<path in this repo>` — adapted from / copied from `<repo>` `<path>` @ `<commit>` (`<license>`); what was taken.
+
+## MPK / mirage (lithos-ai/mirage, branch `low_latency`, Apache-2.0)
+
+- `monolith/registry.py` — adapted from `python/mirage/mpk/models/_registry.py` @ 5beaed8 (Apache-2.0): the
+  duplicate-refusing decorator registry pattern (`register_model`), generalized to every registry kind.
+- `monolith/nn/module.py` — adapted from `python/mirage/mpk/layers_v2/_base.py` @ 5beaed8 (Apache-2.0): the
+  three-method module contract (oracle / lower / weights) and the streaming `load_weights` routing idea; the
+  compile/grid methods, which are CUDA-specific, were not taken.
+
+- `monolith/packs/transforms.py` — `rope_head_perm` adapted from `python/mirage/mpk/models/qwen38/modeling.py` @ 5beaed8
+  (Apache-2.0): the head-dim permutation that implements partial RoPE as a load-time row permutation.
+- `monolith/nn/rope.py` — `rope_tables_permuted` adapted from `python/mirage/mpk/models/qwen38/modeling.py`
+  `build_rope_tables` @ 5beaed8 (Apache-2.0): the cos/sin tables in the permuted head layout.
+- `monolith/models/qwen3_5/config.py` — adapted from `python/mirage/mpk/models/qwen38/configuration.py` @ 5beaed8
+  (Apache-2.0): the field-for-field text-config mirror.
+- `monolith/models/qwen3_5/model.py`, `monolith/models/qwen3_5/weights.py` — adapted from
+  `python/mirage/mpk/models/qwen38/modeling.py` @ 5beaed8 (Apache-2.0): the layer structure, the stacked projection
+  layouts (`[q | gate | k | v]`, `[qkv | z | a | b]`) and the checkpoint key routing; tensor parallelism dropped.
+- `monolith/models/qwen3_5_moe/config.py`, `monolith/models/qwen3_5_moe/model.py` — extend the above
+  hybrid config and model composition with the existing sparse/shared-expert library; retain the same provenance.
+- `tools/goldens/hf_golden.py` — adapted from `tests/runtime_python/models/qwen38/hf_golden.py` @ 5beaed8
+  (Apache-2.0): the golden recipe (BF16, eager attention, hidden states + greedy tokens), extended with
+  teacher-forced top-k logits and a torch-free output format.
+
+## transformers (huggingface/transformers 5.17, Apache-2.0)
+
+* `monolith/nn/rope.py::scaled_inv_freq`: NumPy implementation of the YaRN
+  equations and amplitude scaling in `src/transformers/modeling_rope_utils.py::_compute_yarn_parameters`.
+
+- `monolith/nn/oracle.py`, `monolith/nn/attention.py`, `monolith/nn/gdn.py` — the layer oracles reproduce the
+  semantics and rounding order of `src/transformers/models/qwen3_5/modeling_qwen3_5.py` (`Qwen3_5RMSNorm`,
+  `Qwen3_5RMSNormGated`, `torch_recurrent_gated_delta_rule`, `Qwen3_5GatedDeltaNet`, `Qwen3_5Attention`,
+  `apply_rotary_pos_emb`); written against that file, no code copied verbatim.
+
+- `monolith/models/llama/rope.py` — the frequency-dependent Llama 3 scaling rule
+  follows `src/transformers/modeling_rope_utils.py::_compute_llama3_parameters`
+  @ `v5.17.0` (Apache-2.0), implemented with NumPy for the standalone packer and
+  checked against the installed HF reference. The model package composes the
+  existing layer library; its checkpoint convention is `LlamaForCausalLM`.
+
+## MLX (ml-explore/mlx, MIT)
+
+- `kernels/common/gemv_bf16_small.metal` — adapted from `mlx/backend/metal/kernels/gemv.h`
+  @ `1f8e74e3f12f31365464a6867c6579f0e9b29d85` (MIT): GemvWide's four-vector prefetch/reuse
+  and row reduction, extended for MPK's packed weights, permuted activations, row ranges,
+  residual/statistic and SiLU/multiply epilogues. License: `third_party/MLX-LICENSE`.
+
+- `monolith/formats/nvfp4.py` — `NVFP4_DECODE = 3` in the MSL snippet adapted from `mlx/backend/metal/kernels/fp4.h`
+  (v0.32, MIT): the E2M1 nibble decoded by placing its three magnitude bits into a half's exponent field
+  (`as_type<half>(ushort((bits & 7) << 9))`, the sign a select) with the 2^14 folded into the block scale; the
+  quantized-checkpoint reader also follows MLX's `nvfp4` layout (codes eight per U32, E4M3 scales as U8), no code
+  copied for that.
+  The E4M3 block-scale conversion also adapts `mlx/backend/metal/kernels/fp8.h`
+  @ `1f8e74e3f12f31365464a6867c6579f0e9b29d85`: construct a half from the magnitude
+  bits and apply the exact exponent-bias correction in float, including subnormals.
+  License: `third_party/MLX-LICENSE`.
+
+- `kernels/common/gemv_nvfp4_rows.metal` — adapted from `mlx/backend/metal/kernels/fp_quantized.h`
+  @ `1f8e74e3f12f31365464a6867c6579f0e9b29d85` (MIT): quantization-group dot products
+  reusing four activation vectors across output rows, extended for MPK's packed
+  weights, permuted activations, row ranges and epilogues. License: `third_party/MLX-LICENSE`.
+
+- `kernels/common/gemv_bf16_rows.metal` — adapted from `mlx/backend/metal/kernels/gemv.h`
+  @ `1f8e74e3f12f31365464a6867c6579f0e9b29d85` (MIT): activation-vector reuse across
+  output rows, extended for packed BF16 weights, fused normalization and epilogues.
+  License: `third_party/MLX-LICENSE`.
+
+## Original
+
+- `probes/` — original to this repository.
+
+- `monolith/compiler/moe_fusion.py`, `monolith/compiler/moe_ready.py`, `monolith/compiler/moe_down.py`,
+  `kernels/common/moe_group.metal` — independent
+  Metal implementations informed by Mirage PR #786's expert task grouping and
+  separation of task decomposition from worker placement. Reference:
+  `mirage-project/mirage`, `include/mirage/static_megakernel/expert_queue.cuh`
+  and `python/mirage/mpk/moe.py` at
+  `c28cac618b98fda1e0f1d590923b5f69b4ef3603` (Apache-2.0).
+  No CUDA source copied; no Mirage dependency.
+
+## DeepSpec (deepseek-ai/DeepSpec, MIT)
+
+- `monolith/spec/dspark/model.py`, `tests/spec/test_dspark_oracle.py` — the DSpark drafter's semantics (feature
+  projection, KV injection, bidirectional block attention, the vanilla Markov head, the confidence head, the
+  confident-prefix rule) follow `deepspec/modeling/dspark/qwen3/modeling.py`, `deepspec/modeling/dspark/markov_head.py`
+  and `deepspec/eval/dspark/draft_ops.py` (MIT); written against them, no code copied; the oracle test imports the
+  reference implementation from a local checkout to compare against.
diff --git a/src/inferrsi/data/skill_library/repositories/lithos-metal/scripts/inspect_contract.py b/src/inferrsi/data/skill_library/repositories/lithos-metal/scripts/inspect_contract.py
new file mode 100644
index 000000000..7edc6f743
--- /dev/null
+++ b/src/inferrsi/data/skill_library/repositories/lithos-metal/scripts/inspect_contract.py
@@ -0,0 +1,106 @@
+from __future__ import annotations
+
+import argparse
+import importlib.metadata
+import importlib.util
+import inspect
+import json
+
+
+def inspect_software(expected_version: str, check_contract: bool) -> tuple[dict, bool]:
+    report = {
+        "scope": "installed software/API and optional basic CPU contracts",
+        "backend_runtime_checked": False,
+        "model_numerics_checked": False,
+        "expected_version": expected_version,
+        "checks": [],
+    }
+    try:
+        version = importlib.metadata.version("lithos-metal")
+    except importlib.metadata.PackageNotFoundError:
+        report["checks"].append({"name": "distribution", "ok": False,
+                                  "reason": "lithos-metal is not installed"})
+        return report, False
+    report["distribution_version"] = version
+    try:
+        import monolith
+        from monolith.compiler import compile_program
+        from monolith.core.step_state import StepStateLayout
+        from monolith.generate import Session, load_session
+        from monolith.runtime.prefix_cache import PrefixCache
+    except ImportError as exc:
+        report["checks"].append({"name": "imports", "ok": False,
+                                  "reason": "missing package dependency",
+                                  "exception_type": type(exc).__name__})
+        return report, False
+    report["module_version"] = monolith.__version__
+    report["checks"].append({"name": "version_agreement",
+                              "ok": version == monolith.__version__ == expected_version})
+    report["api_parameters"] = {
+        name: list(inspect.signature(obj).parameters)
+        for name, obj in (
+            ("Session", Session), ("Session.generate", Session.generate),
+            ("load_session", load_session), ("compile_program", compile_program),
+            ("StepStateLayout", StepStateLayout), ("PrefixCache", PrefixCache),
+        )
+    }
+    report["optional_modules_discoverable"] = {
+        name: importlib.util.find_spec(name) is not None
+        for name in ("fastapi", "httpx", "transformers", "safetensors")
+    }
+    if check_contract:
+        from monolith.compiler.barriers import place_barriers
+        from monolith.runtime.program import KernelSpec, OpSpec, Program
+
+        layout = StepStateLayout(t_max=4, gamma_max=3)
+        decoded = layout.unpack(layout.pack({"anchor": -1, "pending_tokens": [11, 12]}))
+        report["checks"].append({
+            "name": "layout_roundtrip_and_alignment",
+            "ok": decoded["anchor"] == -1 and decoded["pending_tokens"] == [11, 12, 0, 0]
+            and layout.size % 16 == 0 and len(set(layout.offsets.values())) == len(layout.fields),
+        })
+        rejected = False
+        try:
+            StepStateLayout(t_max=2, gamma_max=2)
+        except ValueError:
+            rejected = True
+        report["checks"].append({"name": "invalid_layout_rejected", "ok": rejected})
+        program = Program(
+            {"k": KernelSpec("", "k")}, {}, [
+                OpSpec("k", [(0, "x", 0), (1, "a", 0)], (1, 1, 1), (32, 1, 1),
+                       meta={"writes": [1]}),
+                OpSpec("k", [(0, "a", 0), (1, "out", 0)], (1, 1, 1), (32, 1, 1),
+                       meta={"writes": [1]}),
+            ],
+        )
+        count = place_barriers(program)
+        report["checks"].append({
+            "name": "dependent_dispatch_barriers",
+            "ok": count == 2 and all(op.barrier_before for op in program.ops),
+        })
+    return report, all(item["ok"] for item in report["checks"])
+
+
+def main() -> int:
+    parser = argparse.ArgumentParser(
+        description="Inspect installed Lithos Metal software and optional tiny CPU contracts. "
+        "Never uses network, devices, models, subprocesses, file writes or the source checkout."
+    )
+    parser.add_argument("--expected-version", default="0.1.2")
+    parser.add_argument("--check-contract", action="store_true",
+                        help="Run tiny CPU layout and dependency assertions; never create a device")
+    args = parser.parse_args()
+    try:
+        report, ok = inspect_software(args.expected_version, args.check_contract)
+    except Exception as exc:
+        # Exception messages may contain local paths; retain only the failure type.
+        report, ok = {"scope": "software/CPU contract inspection", "ok": False,
+                      "exception_type": type(exc).__name__,
+                      "reason": "inspection failed; inspect the selected package privately"}, False
+    report["ok"] = ok
+    print(json.dumps(report, indent=2, sort_keys=True))
+    return 0 if ok else 1
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())
diff --git a/src/inferrsi/data/skill_library/repositories/lithos-metal/sub-skills/compile-fusion/SKILL.md b/src/inferrsi/data/skill_library/repositories/lithos-metal/sub-skills/compile-fusion/SKILL.md
new file mode 100644
index 000000000..2206f1a02
--- /dev/null
+++ b/src/inferrsi/data/skill_library/repositories/lithos-metal/sub-skills/compile-fusion/SKILL.md
@@ -0,0 +1,65 @@
+---
+name: compile-fusion
+description: "Audit Lithos Metal task compilation, chip-qualified fusion, normalized controls, ICB barriers and scratch lifetimes when changing a recipe or diagnosing a fused kernel."
+---
+
+# Compile and qualify a fusion region
+
+## Trigger and evidence
+
+Use for missing kernel coverage, a recipe rejected by shape/layout checks,
+incorrect fusion, synchronization failures or apparent dispatch reductions.
+Source-derived candidate; empirical validation **not_run**. Read the
+[root](../../SKILL.md) for version scope and the
+[source readset](../../references/source-notes.md) for exact evidence.
+
+## Inputs and owner
+
+Capture target/draft operation graph, packed formats/layouts, active and compiled
+row bounds, context capacity, selected chip/backend/config digest, numerical
+options, region recipe and emitted `Program`. Identify a model-graph change
+separately from a chip schedule change. Do not add model-name conditions to a
+generic compiler to bypass a failing contract.
+
+## Decision procedure
+
+1. Follow model lowering → typed graph checks/passes → coverage guard → emitted
+   kernels/buffers/ops → backend finalization → runtime ICB. Missing bindings
+   fail before execution; register an implementation or retain a supported
+   route instead of suppressing `CoverageError`.
+2. Establish the region boundary. Layer mixer fusion can include projection,
+   mixer core/reduction, output/residual/normalization; MLP and draft/context
+   work can be separate. Read [mechanisms](references/mechanisms.md) to map
+   actual task roles and avoid assuming one kernel spans the model.
+3. Build a dependency ledger for every buffer: producer, consumers, byte/view
+   alias, role and lifetime. `meta['writes']` enables hazard reasoning;
+   missing writes metadata is conservative. Check first-step and inter-step
+   dependencies as well as dependencies within a region.
+4. Compare original/native, normalized multi-dispatch and fused programs when
+   studying fusion itself. Normalization changes loading/layout/geometry, so
+   native-to-fused timing combines effects. Keep decoded weights and rounding
+   mode matched. `commute_norm=False` is a separate numerical control.
+5. Preserve StepState predicates, bounded worker synchronization and recurrent
+   commit passes. Worker count is a logical launch parameter, not a guarantee
+   of physical core placement or simultaneous residency. Do not remove a
+   barrier based only on an observed launch order.
+6. If scratch is reused, prove a known producer and later consumer with disjoint
+   lifetimes. Keep initialized/input/output, persistent state, acceptance logs,
+   ring and parameters out of unsafe aliasing. Recompute hazards after renaming.
+7. Check CPU graph/barrier/arena contracts, then actual Metal equivalence and
+   model continuation/state before benchmarking the complete claimed interval.
+   Report unsupported shapes and unqualified chip variants separately.
+
+## Counter-signals and outputs
+
+A native fallback is expected when an eligible measured recipe is absent.
+Fewer dispatches can lose to synchronization, normalization, occupancy or memory
+cost. A shader compile success does not prove binding bounds or numerics.
+`barriers='all'` is an ordering control, not a cure for unsafe cross-worker
+publication inside one megakernel.
+
+Return a bounded region/dependency ledger, chosen control, eligibility/fallback
+decision and validation gaps. Read [troubleshooting](references/troubleshooting.md)
+for concrete coverage, geometry, barrier and scratch failures. Changing a
+recipe requires its own qualified device evidence; this skill supplies no
+universal tile sizes, performance gates or measured gains.
diff --git a/src/inferrsi/data/skill_library/repositories/lithos-metal/sub-skills/compile-fusion/metadata.json b/src/inferrsi/data/skill_library/repositories/lithos-metal/sub-skills/compile-fusion/metadata.json
new file mode 100644
index 000000000..4a412dfcf
--- /dev/null
+++ b/src/inferrsi/data/skill_library/repositories/lithos-metal/sub-skills/compile-fusion/metadata.json
@@ -0,0 +1,52 @@
+{
+  "profile": "local-skill-v1",
+  "schema_version": 1,
+  "kind": "skill",
+  "origin": [
+    "external"
+  ],
+  "status": "candidate",
+  "empirical_validation": "not_run",
+  "tags": [
+    "lithos-metal",
+    "metal",
+    "apple-silicon",
+    "compile-fusion"
+  ],
+  "sources": [
+    {
+      "type": "git",
+      "url": "https://github.com/lithos-ai/lithos-metal",
+      "commit": "6653b7df5f2c4392ec7b038960f2796b1499923f",
+      "paths": [
+        "monolith/compiler/emit.py",
+        "monolith/compiler/region_fusion.py",
+        "monolith/compiler/static_fusion.py",
+        "monolith/compiler/decoder_fusion.py",
+        "monolith/compiler/barriers.py",
+        "monolith/compiler/arena.py",
+        "monolith/compiler/coverage.py",
+        "monolith/backends/metal/base.py",
+        "monolith/backends/metal/registry.py"
+      ],
+      "readset": "../../references/source-notes.md"
+    }
+  ],
+  "applicability": {
+    "scope": "Pinned Lithos Metal 0.1.2 source/API contracts; Apple silicon Metal runtime; model and recipe qualification required separately."
+  },
+  "limitations": [
+    "Source-derived candidate; empirical efficacy validation not_run.",
+    "CPU/software and synthetic Metal contracts do not qualify full-model/DSpark numerics, quality or E2E performance.",
+    "No model downloads, live DisCo deployment or inference backend integration is provided."
+  ],
+  "provenance_readsets": [
+    "../../references/source-notes.md"
+  ],
+  "consumption_mode": "packaged",
+  "license": "Apache-2.0",
+  "disable-model-invocation": true,
+  "metadata": {
+    "disco-role": "operating"
+  }
+}
diff --git a/src/inferrsi/data/skill_library/repositories/lithos-metal/sub-skills/compile-fusion/references/mechanisms.md b/src/inferrsi/data/skill_library/repositories/lithos-metal/sub-skills/compile-fusion/references/mechanisms.md
new file mode 100644
index 000000000..3462d9ff9
--- /dev/null
+++ b/src/inferrsi/data/skill_library/repositories/lithos-metal/sub-skills/compile-fusion/references/mechanisms.md
@@ -0,0 +1,79 @@
+# Compiler and fusion mechanisms
+
+Use to turn a compile/fusion question into a concrete region and dependency
+inspection. These are pinned implementation facts, not evidence of efficacy.
+
+## Program boundary
+
+`compile_program` lowers a model, checks IR and runs passes. `emit_program`
+checks registered op bindings for the profile before emission. It specializes
+kernels, binds packed weight windows, allocates role-tagged buffers, emits
+operation metadata and places dependency barriers before backend finalization.
+`Program` records `KernelSpec`, `BufferSpec`, `OpSpec`, `StepStateLayout`, context
+capacity and backend identity. ICB arguments live in buffers rather than
+per-replay `setBytes` calls.
+
+The backend owns handler/finalize hooks, decoder/draft/prefill optimization,
+device validation and serving recipes. Device selection matches chip, Apple
+family and core count. Backend cache identity covers configuration fingerprint,
+Metal sources, backend implementation and format decode sources. A tuning cache
+hit does not establish correctness at a different source/device combination.
+
+## Fusion comparison
+
+`region_fusion.fuse_regions` validates disjoint, nonempty, uniquely named spans,
+normalizes each region, repairs layouts/weight windows and returns both the
+normalized control and selected program. `mode='native'` or `fuse=False` keeps
+normalized dispatches. `merge_labeled_regions` merges eligible regions into a
+megakernel and carries written-binding metadata and a StepState done/error guard.
+
+`static_fusion.normalize` can change cooperative loading, tile packing, scratch
+and threadgroup geometry before fusion. `merge` generates fixed-worker stage,
+interleave or queue scheduling with validated geometry. These are distinct
+transformations. The normalized control is necessary to attribute a difference
+specifically to merging.
+
+`decoder_fusion.optimize` discovers mixer and two-projection MLP boundaries from
+emitted kernel roles. Mixer, MLP and vocabulary regions are independently
+selected by recipe; prefill and draft regions remain separate owners. Static
+GDN mixer finalization has narrower preconditions than dynamic speculative
+decoder optimization. A general description of mixer fusion is not permission
+to force one recipe through every emitter path.
+
+## Ordering and storage ledger
+
+| Mechanism | Actual obligation | Useful control/counter-signal |
+| --- | --- | --- |
+| `place_barriers` | Buffer-level RAW/WAR/WAW hazards over an open group; row views alias their base; first op waits for prior step | `minimal` versus `all`; unknown writes metadata is treated as writing all bindings |
+| Mutually exclusive T variants | May share a variant group because only one executes | Prove predicate exclusivity and metadata before grouping new ops |
+| Megakernel local scratch | Threadgroup barriers protect local use | ICB barriers cannot order two tasks inside one dispatch |
+| Cross-worker publication | Device-memory ordering and bounded stage barriers/task queues | Bounded timeout is surfaced through StepState; launch order alone is insufficient |
+| `reuse_arenas` | Known producer, later consumer, nonoverlapping physical lifetimes; preserve offsets and recompute barriers | Unknown producer, read-before-write, initialized storage and observable outputs stay separate |
+| `Engine` shared buffers | Reuse compatible names/sizes except program-owned `params`; reject undersized state/weights/ring | Same name is insufficient for differently packed weight windows during session adoption |
+
+The arena pass preserves StepState, ring, logits, acceptance/confidence logs,
+non-arena roles, initialized/file-backed storage and externally supplied or
+terminal values. Sequential attention partial workspaces can share capacity
+only with dependency barriers ordering reuse; they are not persistent KV state.
+
+## Numerical controls
+
+Hoisting residual-producer RMS statistics preserves the producer's BF16 values
+and specified reduction order. Commuted input normalization is a different
+optimization and intentionally changes BF16 rounding; it is enabled by default
+for eligible projections. Use `Session(..., commute_norm=False)` or the
+generation command's `--no-commute-norm` when establishing a matching control.
+Do not call a changed rounding path bit-exact without a corresponding comparison.
+
+## Proposed verification sequence
+
+1. Compile tiny graphs and inspect missing-op, layout/bounds, writes and arena
+   assertions without allocating a device.
+2. Run synthetic Metal ICB/reencode and state/ring checks on the selected device.
+3. Compare original/normalized/fused region outputs and all bound state with
+   matching decoded weights, rows and numerical options; include boundary rows.
+4. Compare complete continuation and rollback before measuring round time.
+5. For a service claim, include setup/warmup, prefill, drafting, target verify,
+   acceptance/rewind and HTTP/stream completion under the frozen workload.
+
+The sequence is proposed; no measured gain or full-model validation is bundled.
diff --git a/src/inferrsi/data/skill_library/repositories/lithos-metal/sub-skills/compile-fusion/references/troubleshooting.md b/src/inferrsi/data/skill_library/repositories/lithos-metal/sub-skills/compile-fusion/references/troubleshooting.md
new file mode 100644
index 000000000..0e99a99f2
--- /dev/null
+++ b/src/inferrsi/data/skill_library/repositories/lithos-metal/sub-skills/compile-fusion/references/troubleshooting.md
@@ -0,0 +1,16 @@
+# Compilation and fusion failures
+
+| Symptom | Inspect | Next bounded action |
+| --- | --- | --- |
+| `CoverageError` lists ops with no kernel | Profile key, op registry and actual graph | Supply a supported binding or supported model path; never defer the missing op to runtime. |
+| `fusion regions must be disjoint...` | Emitted span endpoints and unique labels after prior transformations | Reconstruct boundaries from current kernel roles; do not reuse stale indices. |
+| Unsupported tile/operand/worker geometry | Format/layout, active row bound, normalization mode and chip recipe | Retain native/normalized control. A recipe copied from another chip is unqualified. |
+| Fused wrong, normalized correct | Written bindings, StepState predicates, local versus cross-worker publication, task bounds | Reduce to one region with the same numerical control. Preserve a bounded failure; do not lengthen spin loops blindly. |
+| `Megakernel: bounded worker barrier timed out` / state error 3 | Stage participation, uniform barriers, task queue bounds and device/recipe match | Stop the affected experiment; compare safe native path and capture minimal diagnostic state. `barriers='all'` does not fix intrakernel deadlock. |
+| Incorrect only after scratch reuse | Known producers, first reads, preserved outputs/state, offsets and renamed hazards | Retain original buffers and compare. Reapply barrier placement after any physical alias change. |
+| Dispatch count drops but round slows | Original vs normalized vs fused cost, sync, spills/storage and entire region | Keep all observations; reject a speedup inference from dispatch count alone. |
+| Numerical drift after normalization | `commute_norm`, decoded format, fast math, reference rounding/reduction order | Match numerical settings before attributing a fusion defect; qualify intentional changed rounding independently. |
+
+A workaround is a candidate change until checked against actual selected rows,
+state and backend. Source tests and isolated timings cannot establish service
+quality or performance.
diff --git a/src/inferrsi/data/skill_library/repositories/lithos-metal/sub-skills/generation-state/SKILL.md b/src/inferrsi/data/skill_library/repositories/lithos-metal/sub-skills/generation-state/SKILL.md
new file mode 100644
index 000000000..bd314f773
--- /dev/null
+++ b/src/inferrsi/data/skill_library/repositories/lithos-metal/sub-skills/generation-state/SKILL.md
@@ -0,0 +1,68 @@
+---
+name: generation-state
+description: "Trace Lithos Metal GPU-resident generation and DSpark verification, commit/rollback, token-ring and cancellation state when continuation, acceptance or stop behavior is wrong."
+---
+
+# Trace generation and speculative state
+
+## Trigger and prerequisites
+
+Use for continuation drift, wrong tokens after a rejected proposal, recurrent
+state mismatch, context/ring errors, unexpected stop or cancellation latency.
+Source-derived candidate; empirical validation **not_run**. Read the
+[root](../../SKILL.md), [state mechanism](references/state-machine.md) and
+[source readset](../../references/source-notes.md).
+
+Record target/draft identities, tokenizer, model/DSpark configuration, taps,
+formats/precision, context capacities, layout bounds, verification and sampling
+policy, numerical options and selected backend/recipe. Dimensions alone do not
+prove target/draft compatibility. Server DSpark uses fixed verification;
+lower-level `Session` defaults to cost-aware selection with a fallback when
+no matching cost table is available.
+
+## Decision procedure
+
+1. Identify whether the failing pass is intermediate prefill, final prefill,
+   target verification, recurrent commit, draft context injection, proposal
+   construction, verify-length selection or host token draining. Do not treat
+   every replay as a productive decode round.
+2. Reconstruct the committed-prefix ledger across the failing boundary. Include
+   `position`, `t_this_step`, `pending_tokens`, `verify_len`, `accepted`,
+   `checkpoint_index`, `n_inject`, `drafter_ctx_len`, recurrent slots,
+   `ring_head/tail`, `stop_at`, `done` and `error`. Match the shared ABI rather
+   than hardcoding offsets from another layout.
+3. For `L` selected proposals, account for `1 + L` target rows including the
+   anchor. Acceptance commits a prefix plus correction/bonus, bounded by EOS.
+   Rejected positions must not survive as logical target or draft history.
+   Inspect convolution and GDN commit passes as well as attention lengths;
+   hidden-output agreement alone is insufficient.
+4. Distinguish greedy/argmax-proposal and sampled-proposal semantics. The sampled
+   path retains `q` logits/statistics through verification and uses rejection
+   correction. Never apply its probability rule to the argmax proposal path or
+   persist per-round proposal scratch as a prefix checkpoint.
+5. Check compiled/active row and context bounds before varying proposal count.
+   A changed block can select another recipe or exceed layout capacity. Preserve
+   target/draft configuration and sampling semantics in any target-only control.
+6. For streaming/cancellation, trace completed GPU work and committed ring tokens.
+   Callbacks receive cumulative verified output. Cancellation is cooperative
+   between bounded host pumps; an active Metal dispatch completes first.
+   Never free buffers still owned by in-flight work.
+7. Propose a bounded reproduction covering zero/full/interior rejection, EOS,
+   token limit, context end and replay after done. Compare continuation tokens
+   and target/draft persistent state with matched numerical mode. CPU mocks and
+   synthetic ICB checks are useful contracts; actual DSpark/model numerics still
+   require the selected checkpoint/backend.
+
+## Counter-signals and report
+
+Queued replays may return immediately after `done`; submitted steps can exceed
+productive rounds. Full acceptance still needs correction/bonus accounting.
+Allocated KV rows past the committed length need not be physically erased if
+they are never read as valid history. An expected cache miss or conservative
+row fallback is not evidence of corrupted state.
+
+Return the first mismatched state obligation, its owner, the exact failing
+round/prefix, control configuration and missing observations. Read
+[troubleshooting](references/troubleshooting.md) for ABI, ring, capacity and
+state-recovery failures. Do not publish complete-round or service rates from
+target-only compute projections or acceptance assumptions.
diff --git a/src/inferrsi/data/skill_library/repositories/lithos-metal/sub-skills/generation-state/metadata.json b/src/inferrsi/data/skill_library/repositories/lithos-metal/sub-skills/generation-state/metadata.json
new file mode 100644
index 000000000..921695d95
--- /dev/null
+++ b/src/inferrsi/data/skill_library/repositories/lithos-metal/sub-skills/generation-state/metadata.json
@@ -0,0 +1,49 @@
+{
+  "profile": "local-skill-v1",
+  "schema_version": 1,
+  "kind": "skill",
+  "origin": [
+    "external"
+  ],
+  "status": "candidate",
+  "empirical_validation": "not_run",
+  "tags": [
+    "lithos-metal",
+    "metal",
+    "apple-silicon",
+    "generation-state"
+  ],
+  "sources": [
+    {
+      "type": "git",
+      "url": "https://github.com/lithos-ai/lithos-metal",
+      "commit": "6653b7df5f2c4392ec7b038960f2796b1499923f",
+      "paths": [
+        "monolith/core/step_state.py",
+        "monolith/generate.py",
+        "monolith/spec/dspark/model.py",
+        "monolith/runtime/engine.py",
+        "runtime/src/metal_core.mm",
+        "kernels/common/spec_ops.metal"
+      ],
+      "readset": "../../references/source-notes.md"
+    }
+  ],
+  "applicability": {
+    "scope": "Pinned Lithos Metal 0.1.2 source/API contracts; Apple silicon Metal runtime; model and recipe qualification required separately."
+  },
+  "limitations": [
+    "Source-derived candidate; empirical efficacy validation not_run.",
+    "CPU/software and synthetic Metal contracts do not qualify full-model/DSpark numerics, quality or E2E performance.",
+    "No model downloads, live DisCo deployment or inference backend integration is provided."
+  ],
+  "provenance_readsets": [
+    "../../references/source-notes.md"
+  ],
+  "consumption_mode": "packaged",
+  "license": "Apache-2.0",
+  "disable-model-invocation": true,
+  "metadata": {
+    "disco-role": "operating"
+  }
+}
diff --git a/src/inferrsi/data/skill_library/repositories/lithos-metal/sub-skills/generation-state/references/state-machine.md b/src/inferrsi/data/skill_library/repositories/lithos-metal/sub-skills/generation-state/references/state-machine.md
new file mode 100644
index 000000000..982136d0b
--- /dev/null
+++ b/src/inferrsi/data/skill_library/repositories/lithos-metal/sub-skills/generation-state/references/state-machine.md
@@ -0,0 +1,100 @@
+# Generation and DSpark state contracts
+
+Use to construct a round-by-round ledger without needing the original checkout.
+The implementation facts below are scoped to the pinned source; verification
+steps are proposals, not bundled model evidence.
+
+## Shared control ABI and storage
+
+`StepStateLayout(t_max, gamma_max)` requires room for the anchor plus maximum
+draft block. Fields are four-byte scalars/arrays, total storage is padded to
+16 bytes, and order is a compiler/runtime ABI. Fields are appended rather than
+reordered. `pack`/`unpack` are little-endian host representations; emitted MSL
+uses the same offsets. The root software helper checks a tiny roundtrip and
+invalid-bound rejection; it never exercises GPU rollback.
+
+| State | Meaning/owner |
+| --- | --- |
+| `position`, active rows and `pending_tokens` | First row position and inputs of current replay; intermediate prompt chunks are host-fed |
+| `anchor`, `gamma`, proposal/confidence arrays, `verify_len` | Draft/selection state; target verification consumes anchor plus selected proposals |
+| `accepted`, `checkpoint_index`, `n_inject` | Accepted prefix and recurrent commit slot/rows; committed target features for next draft pass |
+| `drafter_ctx_len`, draft K/V | Injected committed context, distinct from proposals being verified |
+| `rng_lo/hi` | Sampling reproducibility; seed alone is insufficient if policy/round order differs |
+| `ring_head/tail`, `stop_at`, `done`, `error` | GPU publication, host consumption, request stop and bounded failure |
+
+Weights are immutable packed mappings. Persistent target K/V, convolution/GDN
+slots, draft context and control survive replays; acceptance logs and ring have
+their own lifetimes. Temporary activations/partials may be reused only after
+their last consumer. Program parameter records belong to a particular program.
+
+## Speculative graph and commit
+
+The compiler's `lower_round` appends acceptance to the target graph, then every
+operation's declared `commit_kind`, then target feature taps → drafter block →
+verify selection. The DSpark model injects committed features in its next draft
+pass; its `lower_context_update` is deliberately a no-op, not missing work.
+
+1. Intermediate prefill has `prefill_left > 0`; acceptance advances by prompt
+   rows, selects the matching recurrent checkpoint, sets injection rows and
+   emits no output. The final prompt pass samples the first token.
+2. `verify_select` advances injected-context bookkeeping, bounds selected `L`
+   by block, layout and remaining context, and writes pending anchor/proposals
+   and `t_this_step = 1 + L` for the next target replay.
+3. `accept_scan` finds the matching proposal prefix, obtains a target correction
+   or bonus, checks ring space, publishes sequence-tagged tokens through the
+   first EOS, and updates position/anchor/commit bookkeeping. For a first pass
+   with prompt rows, `base = t_this_step - 1 - verify_len` accounts for rows
+   preceding the anchor. Do not derive committed state from `accepted` alone.
+4. Convolution/recurrent commit paths retain/recompute the state matching those
+   committed rows. Attention validity follows logical positions; overwritten
+   rejected rows cannot be included by later length bookkeeping.
+5. The request's output cap can stop after a complete verified round. The host
+   truncates returned/callback tokens to the requested output limit; queued
+   dispatches after `done` return without productive work. Use acceptance logs
+   for productive speculative round counts rather than all submitted steps.
+
+## Sampling and verification policies
+
+Greedy acceptance compares target argmax with proposals. For nonzero temperature,
+the serving default still uses argmax draft proposals: accept when they equal
+the target sample. Sampled proposals use draft distribution `q`, acceptance
+`min(1, p/q)` and positive-part `p-q` replacement at rejection. Target top-k
+renormalization precedes top-p on that path. The sampled path requires proposal
+logits/statistics; those per-round values are excluded from persistent prefix
+snapshots. Changing draft precision can change proposals and acceptance.
+
+Lower-level selection supports `fixed`, `threshold` and `cost`. Fixed length
+must be within the drafter block. Cost uses a backend cost table for the relevant
+pack format; absence falls back to threshold, not extrapolated cost. Serving
+selects fixed verification up to the supported block, normally seven proposals
+plus anchor. A confidence head does not prove confidence scheduling is active.
+
+DSpark's block transformer uses target feature K/V and bidirectional block
+attention; sequential Markov corrections restore predecessor dependence and a
+confidence head supplies policy estimates. Check checkpoint-owned embedding/head
+versus supported target sharing. `--draft-quantization none` preserves supplied
+precision; it cannot recreate BF16 from an already quantized source.
+
+## Stop, failure and host ownership
+
+The ring holds aligned eight-byte `(sequence << 32) | token` slots, so wrapped
+slots can be distinguished. Ring capacity bounds publication and host draining.
+State error 1 is ring overflow, 2 context capacity and 3 bounded mixer barrier
+timeout; `Engine.run` explicitly raises for error 3. `Session.generate` raises
+other stopped-short failures and rejects over-capacity requests before work.
+An error reached only by queued work after a fully served request must be
+interpreted with actual committed output, not the raw flag alone.
+
+`Engine.run` waits for its queued work to finish. Streaming pumps are bounded;
+cancel predicates stop subsequent work after the active pump. The interface
+does not promise preemption of an active kernel. Preserve lifetime through that
+completion boundary before release/adoption or a new request reset.
+
+## Proposed qualification
+
+Start with binary-layout and mock generation tests; then exercise real synthetic
+ICB/reencode, done, wrap and overflow. Full-model tests must compare deterministic
+continuation, rejection boundary, recurrent slots and target/draft logical
+lengths; stochastic tests need the configured distribution protocol. Include
+prefill and cache-resume boundaries. A hidden tensor comparison, accepted-count
+summary or target-only latency is insufficient by itself.
diff --git a/src/inferrsi/data/skill_library/repositories/lithos-metal/sub-skills/generation-state/references/troubleshooting.md b/src/inferrsi/data/skill_library/repositories/lithos-metal/sub-skills/generation-state/references/troubleshooting.md
new file mode 100644
index 000000000..4de904212
--- /dev/null
+++ b/src/inferrsi/data/skill_library/repositories/lithos-metal/sub-skills/generation-state/references/troubleshooting.md
@@ -0,0 +1,17 @@
+# Generation-state failures
+
+| Symptom | Discriminating evidence | Recovery/check |
+| --- | --- | --- |
+| Invalid `StepStateLayout` or wrong field values | `t_max`, `gamma_max`, compiled layout size/offsets and serialized byte count | Rebuild the program with a compatible shared layout; never patch one side's offsets alone. |
+| Speculative compile rejects a block | Dynamic-T flag, draft pack, gamma versus layout and verify length | Supply compatible bounds and policy. A smaller runtime `L` cannot repair an undersized compiled allocation. |
+| Correct hidden rows but wrong continuation after rejection | Accepted prefix plus correction, checkpoint index, GDN/conv state, draft context length | Compare the first rejected round's complete persistent state. Resetting only target K/V misses recurrent/draft owners. |
+| Unexpected acceptance differences | Argmax versus sampled proposals, target sampling, draft precision/head and feature taps | Match policies/checkpoints before blaming a kernel. Never reinterpret confidence scheduling as the fixed serving default. |
+| Ring overflow / error 1 | Capacity, produced batch, head/tail sequences and host drain cadence | Reproduce with tiny synthetic ring; review bounded pump/capacity choices separately. Do not discard overflowed tokens and claim correctness. |
+| Capacity reached / error 2 | Prompt plus output, target/draft capacities and reserved block rows | Reduce within an authorized context or rebuild compatible packs/programs; distinguish stopped-short output from queued no-op overrun. |
+| Stop seems late | Submitted steps versus productive acceptance logs, callback cap and in-flight pumps | Inspect committed output and `done`; cancellation does not preempt an active Metal dispatch. |
+| Repeated streaming step publishes nothing | `done`, state error and ring progress | Preserve `Streaming decode made no progress` as failure; avoid an unbounded retry loop. |
+| Cache-resumed output differs | Exact tokens, saved recurrent parity, StepState and target/draft buffers | Use the serving-cache route to inspect snapshot integrity/identity before disabling generation safeguards. |
+
+Unknown target/draft pairings, unavailable checkpoints/oracles, unsupported row
+recipes or unobserved state remain explicit limits. Do not count CPU mocks as
+DSpark numerical validation or publish target-compute projections as E2E rates.
diff --git a/src/inferrsi/data/skill_library/repositories/lithos-metal/sub-skills/serving-cache/SKILL.md b/src/inferrsi/data/skill_library/repositories/lithos-metal/sub-skills/serving-cache/SKILL.md
new file mode 100644
index 000000000..104c045de
--- /dev/null
+++ b/src/inferrsi/data/skill_library/repositories/lithos-metal/sub-skills/serving-cache/SKILL.md
@@ -0,0 +1,66 @@
+---
+name: serving-cache
+description: "Audit Lithos Metal local serving, prepared-pack identity, exact target/draft prefix checkpoints, disk integrity, session adoption and idle release when startup, cache reuse or request isolation fails."
+---
+
+# Validate serving and cache ownership
+
+## Trigger and evidence
+
+Use for stale pack/prefix reuse, slow or failed startup, cache-resume drift,
+HTTP state isolation, session replacement and idle reload. Source-derived
+candidate; empirical validation **not_run**. Read the [root](../../SKILL.md),
+[workflows and contracts](references/workflows.md) and
+[source readset](../../references/source-notes.md).
+
+Obtain selected target/draft revisions, checkpoint/pack manifests, tokenizer/chat
+template identity, backend/config/recipe, context/prefill/numerical settings,
+actual prompt token boundaries, tier/budget and load/unload/request trace.
+Use placeholders for local paths and private prompt content in reports.
+
+## Decision procedure
+
+1. Separate checkpoint resolution, derived weight packs, CPU programs/pipelines,
+   GPU session buffers and host/disk prefix snapshots. Each has a different
+   validity key and lifetime. A pack hit is not a prefix hit or a model check.
+2. Validate local assets and prepared packs before serving. Check architecture,
+   all shards, context capacity, layouts/formats, selected quantization and
+   manifest ranges. Preserve a reported mismatch; do not reuse or repair an
+   incompatible cache silently. File size/mtime identity is not weight-byte
+   authentication.
+3. Trace tokenizer/template input through system messages, tools and history to
+   exact token IDs. Prefix matching uses the longest **strict** matching prefix,
+   leaving at least one input token for final prefill/logits. Equivalent text,
+   a changed tool schema or an exact-length request is not automatically a hit.
+4. Check a snapshot's StepState, live target/draft attention rows and complete
+   recurrent slots before any restore writes. Worker counters and per-round
+   proposal scratch are not sequence checkpoints. For disk, also check identity,
+   lengths/names/ranges and per-blob hashes; incompatible/corrupt entries fall
+   back to prefill.
+5. On session/recipe/sampling changes, inspect buffer adoption and state reset.
+   Only selected GPU buffers remain resident; CPU programs and executable
+   pipelines can survive. Same-named differently packed weights cannot be
+   shared blindly. Distinguish reuse of allocation from reuse of sequence state.
+6. Trace pending arrival → GPU lock → load/generation/stream → completion before
+   idle release. Unload waits for active GPU work, releases engines/GPU
+   allocations, attempts a bounded flush of optional disk copies, then clears
+   host snapshots.
+   A stalled writer can finish later; release does not imply every entry was
+   persisted. Reload can reuse code/pipelines and disk prefixes independently.
+7. Check HTTP contracts with mocks and tiny temporary cache fixtures first.
+   Then qualify actual model cache resume, reload and stream cancellation on
+   the selected backend before measuring complete-request latency or stability.
+
+## Counter-signals and outputs
+
+Cache misses after a changed identity, strict-prefix exclusion or quota eviction
+are expected. Prefix-store lock contention can leave a server using memory-only
+snapshots. Health liveness, warmup completion, pack validity and numerical
+continuation are separate observations. Unified memory does not make host
+snapshots, GPU allocations and OS file-cache pages interchangeable accounting.
+
+Return a cache/lifecycle ownership ledger, hit-or-miss reason, earliest invalid
+boundary, safe fallback and unresolved qualification. Read
+[troubleshooting](references/troubleshooting.md) before deleting anything or
+changing capacity. No live server, downloads, cache deletion or network exposure
+is authorized by reading this procedure.
diff --git a/src/inferrsi/data/skill_library/repositories/lithos-metal/sub-skills/serving-cache/metadata.json b/src/inferrsi/data/skill_library/repositories/lithos-metal/sub-skills/serving-cache/metadata.json
new file mode 100644
index 000000000..9868575ca
--- /dev/null
+++ b/src/inferrsi/data/skill_library/repositories/lithos-metal/sub-skills/serving-cache/metadata.json
@@ -0,0 +1,50 @@
+{
+  "profile": "local-skill-v1",
+  "schema_version": 1,
+  "kind": "skill",
+  "origin": [
+    "external"
+  ],
+  "status": "candidate",
+  "empirical_validation": "not_run",
+  "tags": [
+    "lithos-metal",
+    "metal",
+    "apple-silicon",
+    "serving-cache"
+  ],
+  "sources": [
+    {
+      "type": "git",
+      "url": "https://github.com/lithos-ai/lithos-metal",
+      "commit": "6653b7df5f2c4392ec7b038960f2796b1499923f",
+      "paths": [
+        "monolith/serving/cache.py",
+        "monolith/serving/setup.py",
+        "monolith/runtime/prefix_cache.py",
+        "monolith/runtime/prefix_store.py",
+        "monolith/serving/lifecycle.py",
+        "monolith/serve.py",
+        "monolith/generate.py"
+      ],
+      "readset": "../../references/source-notes.md"
+    }
+  ],
+  "applicability": {
+    "scope": "Pinned Lithos Metal 0.1.2 source/API contracts; Apple silicon Metal runtime; model and recipe qualification required separately."
+  },
+  "limitations": [
+    "Source-derived candidate; empirical efficacy validation not_run.",
+    "CPU/software and synthetic Metal contracts do not qualify full-model/DSpark numerics, quality or E2E performance.",
+    "No model downloads, live DisCo deployment or inference backend integration is provided."
+  ],
+  "provenance_readsets": [
+    "../../references/source-notes.md"
+  ],
+  "consumption_mode": "packaged",
+  "license": "Apache-2.0",
+  "disable-model-invocation": true,
+  "metadata": {
+    "disco-role": "operating"
+  }
+}
diff --git a/src/inferrsi/data/skill_library/repositories/lithos-metal/sub-skills/serving-cache/references/troubleshooting.md b/src/inferrsi/data/skill_library/repositories/lithos-metal/sub-skills/serving-cache/references/troubleshooting.md
new file mode 100644
index 000000000..e7610cc1e
--- /dev/null
+++ b/src/inferrsi/data/skill_library/repositories/lithos-metal/sub-skills/serving-cache/references/troubleshooting.md
@@ -0,0 +1,18 @@
+# Serving and cache failures
+
+| Symptom | Inspect | Recovery or boundary |
+| --- | --- | --- |
+| Missing `config.json`/weight shard | Explicit local path, index and all named safetensors shards | Repair the selected local asset set outside this diagnostic. Do not fall through to network download. |
+| `Incomplete weights.pack`, capacity, format or identity mismatch | Manifest/payload size, requested capacity, layout/backend, target/draft role and quantization | Select a valid compatible cache or separately rebuild into a new derived entry; do not weaken validation or delete live data. |
+| Unexpected prefix miss | Exact token IDs including system/tool template; strict length; `min_tokens`, quota and disk identity | Expected miss should prefill. Similar text or equal-length input is not sufficient for a reusable prefix. |
+| Stored entry drops / `StalePrefix` | Header/blob integrity, buffer names/sizes, StepState ABI and device/code/numerics identity | Preserve safe fallback to miss/shorter prefix before any GPU writes; compare outputs after prefill. |
+| Disk tier disabled / `StoreBusy` | Single owner, private directory, quota and I/O error | Keep documented memory-only operation. Do not remove another process's lock or broaden directory permissions. |
+| Prefix file appears queued but not durable | Submit result, writer completion, `flush` result and `contains` | Queue admission is not persistence. Bounded unload can release while queued writers finish. |
+| Drift after recipe/sampling change | Adoption roles/sizes, exact weight windows, state reset and shared prefix identity | Compare with a clean compatible session. Pipeline identity is separate from sequence state. |
+| HTTP 429 / queued Ollama request | Adapter admission path and current generation/load | Expected one-generation policy; do not advertise batching/concurrency. |
+| HTTP 503 `model_load_failed` | Lifecycle state, failed attempt and waiting arrivals | Report failed load distinctly; later requests may retry. `/health` liveness is not loaded readiness. |
+| Unload seems delayed / memory remains | Pending requests, GPU lock/in-flight work, engine references, host snapshots, CPU pipelines and OS file cache | Respect completion/lock boundary. Different memory owners survive differently; no forced teardown of active work. |
+
+Full-model prefix resume, DSpark state, quality and complete HTTP performance
+remain unverified until separately exercised with the selected assets/backend.
+CPU/mock and disk-integrity checks do not clear those scopes.
diff --git a/src/inferrsi/data/skill_library/repositories/lithos-metal/sub-skills/serving-cache/references/workflows.md b/src/inferrsi/data/skill_library/repositories/lithos-metal/sub-skills/serving-cache/references/workflows.md
new file mode 100644
index 000000000..23b53c171
--- /dev/null
+++ b/src/inferrsi/data/skill_library/repositories/lithos-metal/sub-skills/serving-cache/references/workflows.md
@@ -0,0 +1,124 @@
+# Serving, cache and release workflows
+
+Use for constructing local-only commands and tracing validity/lifetime owners.
+This is a distilled procedure; original source scripts are provenance rather
+than runtime dependencies of this skill.
+
+## Local-only startup planning
+
+Inspect `lithos-metal serve --help` first. A separately authorized local run can
+use the following shape, with existing compatible model assets and writable
+derived-cache locations:
+
+```bash
+lithos-metal serve --model <MODEL_DIR> --no-draft --local-files-only --pack <PACK_CACHE_DIR>
+```
+
+`--no-draft` intentionally selects a target-only control. For existing compatible
+DSpark assets, replace it with `--draft <DRAFT_DIR> --draft-pack <DRAFT_PACK_DIR>`;
+record target/draft revisions and precision. Do not infer automatic pairing for
+an unknown or renamed local copy. `--local-files-only` prevents Hub fetching but
+does not validate missing assets or suppress pack writes, native compilation,
+allocation or warmup. A missing local path must not be silently interpreted as
+permission to download. Add no model-download command to a diagnostic task.
+
+The server default binds loopback; retain that boundary unless exposure is
+separately authorized. Chat Completions, Responses, Anthropic Messages and
+Ollama adapters share generation, but protocol support is bounded: Responses
+is stateless and tools execute in the client. Ordinary concurrent generation
+is rejected with 429; Ollama chat requests wait in FIFO order. `/health` is
+liveness and may report lifecycle state, while `/v1/models` reports an alias.
+Neither endpoint verifies output quality or performance.
+
+## Pack identity and publication
+
+`resolve_checkpoint` checks local config and weight shards or resolves a Hub
+snapshot with the requested revision/local-only policy. `checkpoint_identity`
+records resolved path, config/index hashes, and shard names/sizes/mtime. It
+rejects missing shards and unsafe shard paths. It does **not** hash every weight
+byte, so a frozen model-integrity manifest is a separate qualification input.
+
+`ensure_pack` keys automatic storage by cache version, checkpoint identity,
+capacity, layout, backend, target/draft role and quantization. Draft-head
+ownership also affects the key. A per-key lock coordinates builders; a temporary
+directory is validated and atomically renamed only after checking the checkpoint
+did not change during packing. Explicit manually selected packs are validated
+but may lack full automatic cache identity; do not invent one.
+
+`validate_pack` checks supported manifest version, payload size, sufficient
+capacity, quantization/Markov preservation, optional identity equality, expected
+slab shapes/formats and tensor ranges. This is structural/config validation,
+not a cryptographic hash of the entire packed payload or model output check.
+
+## Exact checkpoint contents and identity
+
+`PrefixCache` copies declared model/drafter entries whose buffers have role
+`state`, plus StepState. Position-major target K/V and draft `k_ctx/v_ctx` retain
+only live prefix rows; convolution/recurrent entries retain complete checkpoint
+slots, including parity encoded by StepState. Worker/barrier flags are excluded
+because they are program state rather than sequence history. Intermediate
+prefill snapshots contain committed prompt state, not pending draft decisions.
+
+The host tier evicts before copying to bound its declared buffer-byte budget.
+It retains at most two newest useful checkpoints, favoring a shared stable
+prefix when refreshing continuation. Those implementation counts are not
+universal tuning recommendations. Snapshot cost can exceed replay benefit on
+short prompts; `min_tokens` and actual copy/replay costs matter.
+
+With a store, a longer strict prefix on disk can be loaded, verified and promoted
+over the memory match. Reused evictions can spill; `flush` attempts every host
+copy and reports whether all are present after bounded waiting. `clear` only
+drops host snapshots, not disk files.
+
+`Session.prefix_identity` includes snapshot format/version/code digest, OS/device,
+backend, target/draft manifests and payload size/mtime, module config, layout,
+state entries, prefill/numerical choices and caller extras. Serving supplies
+recipes and tokenizer/template bytes as extras. Capacity is omitted from the
+position-major state shapes to permit compatible capacities; restore still
+checks allocations and exact state-buffer names/StepState size. Payload
+size/mtime is not payload-byte authentication.
+
+## Disk tier and failure recovery
+
+`PrefixStore` uses one file per entry: magic/header, token IDs, StepState and
+named state blobs with aligned offsets and SHA-256. It holds an exclusive owner
+lock, validates private directory ownership/permissions and uses a bounded
+writer queue. Writes use a temporary file, fsync, atomic replacement and
+directory fsync. LRU quota eviction excludes pinned readers/writers. These
+implementation measures do not by themselves demonstrate recovery under every
+filesystem or power-loss scenario.
+
+Load verifies identity, exact tokens, schema/size/ranges and every blob digest.
+Corruption or I/O failure drops the entry and returns a cache miss. Restore
+checks all buffers before writes. A disk checkpoint that no longer fits raises
+`StalePrefix`, is dropped from both tiers, and generation retries prefill with
+a miss or shorter prefix. Do not hand-copy partial state into live buffers.
+
+## Session and idle ownership
+
+`Backend.select_session` keys CPU sessions by greedy-vs-sampling and recipe.
+One sampling program handles parameter updates; pipelines are shared on the
+same device. Adoption settles prior mapping work, reuses compatible roles/sizes,
+preserves only identical weight-file windows, and clears non-KV state. Old KV
+rows are invalidated logically and overwritten before use. Only the selected
+session keeps GPU buffers, while exact-prefix cache ownership is shared.
+
+Idle lifecycle counts requests from arrival through waiting, generation and
+stream completion/cancellation. Load/unload uses the GPU lock and rechecks
+pending requests after waiting. A failed load releases partial allocation and
+returns `model_load_failed`; arrivals sharing that failed attempt do not each
+start another load, while a later request may retry.
+
+Unload releases engines/allocations and attempts bounded prefix flush before
+clearing host copies. CPU programs/pipelines/tokenizer and OS pack-file cache can
+survive. An unavailable/busy disk tier leaves memory-only operation. A warning
+about unfinished prefix persistence is not a completed durable write.
+
+## Proposed checks
+
+Use tiny mock HTTP/state tests for validation/admission/streaming, temporary
+prefix files for identity/corruption/locks/quota, and a fake clock/lock for idle
+ordering. Then compare actual uncached, memory-hit, disk-hit and post-reload
+continuations and persistent state. Distinguish checkpoint copy/read/reset,
+prefill, decoder and complete HTTP times; only the full claimed workload can
+justify service performance wording.
diff --git a/src/inferrsi/data/skill_library/skill-index.json b/src/inferrsi/data/skill_library/skill-index.json
index e683e9fc6..ec051e1b3 100644
--- a/src/inferrsi/data/skill_library/skill-index.json
+++ b/src/inferrsi/data/skill_library/skill-index.json
@@ -304,6 +304,10 @@
     "weight-loading-optimization": {
       "path": "methods/weight-loading-optimization",
       "backend": "common"
+    },
+    "lithos-metal": {
+      "path": "repositories/lithos-metal",
+      "backend": "common"
     }
   }
 }
diff --git a/tests/check_installed.py b/tests/check_installed.py
index 064c2dc4d..de58c3e09 100644
--- a/tests/check_installed.py
+++ b/tests/check_installed.py
@@ -1,5 +1,3 @@
-"""Run outside the checkout after a noneditable [skills] install, PYTHONPATH unset."""
-
 import contextlib
 import hashlib
 import io
@@ -17,7 +15,7 @@
     "This check requires an installed package"
 )
 records = knowledge_records()
-assert len(records) == 124, "Knowledge is missing from the installation"
+assert len(records) == 125, "Knowledge is missing from the installation"
 assert (
     files("inferrsi")
     .joinpath("data", "repo_packs", "vllm_omni", "knowledge", "README.md")
@@ -42,7 +40,7 @@
         "package:" + record["id"], mlx_document, revision=bundle.revision
     ) == files("inferrsi").joinpath(*mlx_document.split("/")).read_bytes()
 skill_index = search_knowledge(repo="inferrsi", limit=100)
-assert len(skill_index) == 76, "Missing installed skill abstracts"
+assert len(skill_index) == 77, "Missing installed skill abstracts"
 assert all(record["index_type"] == "packaged-skill" for record in skill_index)
 indexed = KnowledgeReader().read("inferrsi.ascend.skills.ascend_operator_profiling")
 assert "data/skill_library/hardware/ascend/ascend-operator-profiling/SKILL.md" in indexed.content
@@ -156,6 +154,17 @@
     assert actual == rlt_source_hashes[upstream]
 
 assert "yaml" not in sys.modules, "Installed metadata reads must not require the skills extra"
+lithos = KnowledgeReader().read("inferrsi.common.skills.lithos_metal")
+lithos_owner = "data/skill_library/repositories/lithos-metal/"
+for relative in (
+    "LICENSE", "references/upstream-notice.md", "references/mlx-license.md",
+    "references/source-notes.md", "scripts/inspect_contract.py",
+    "sub-skills/compile-fusion/SKILL.md", "sub-skills/generation-state/SKILL.md",
+    "sub-skills/serving-cache/SKILL.md",
+):
+    assert lithos_owner + relative in lithos.content, relative
+assert json.loads(lithos.content["record.json"])["skill_metadata"]["status"] == "candidate"
+assert "monolith" not in sys.modules, "Skill reads must not import the source engine"
 selection_reader = KnowledgeReader()
 selection = selection_reader.collect(
     "cpu.stat.local", relevance="cgroupv2", context_bytes=65536, limit=1,
@@ -190,7 +199,7 @@
         data = resource.read_bytes().replace(b"\r\n", b"\n")
         assert hashlib.sha256(data).hexdigest() == expected_hash, relative
 registry = json.loads(library.joinpath("skill-index.json").read_text(encoding="utf-8"))["skills"]
-assert len(registry) == 76
+assert len(registry) == 77
 for entry in registry.values():
     assert library.joinpath(*entry["path"].split("/"), "SKILL.md").is_file()
 router = library.joinpath("workflows", "inference-engine-research")
diff --git a/tests/test_knowledge_contract.py b/tests/test_knowledge_contract.py
index e3599f403..301cf180a 100644
--- a/tests/test_knowledge_contract.py
+++ b/tests/test_knowledge_contract.py
@@ -31,7 +31,7 @@ def test_local_profiles_preserve_core_and_structured_ownership():
         assert metadata["kind"] == "skill"
         for readset in metadata["provenance_readsets"]:
             assert (sidecar.parent / readset).is_file()
-    assert len(normalized) == 60
+    assert len(normalized) == 61
 
 
 def test_local_sidecars_omit_fields_that_equal_reader_defaults():
diff --git a/tests/test_skill_index.py b/tests/test_skill_index.py
index 3df1f349f..1802cb810 100644
--- a/tests/test_skill_index.py
+++ b/tests/test_skill_index.py
@@ -1,5 +1,3 @@
-"""The installed index preserves categorized skills and their complete resources."""
-
 import hashlib
 import json
 import subprocess
@@ -23,7 +21,7 @@ def test_complete_index_retains_metadata_states_and_unknown_license():
     records = search_knowledge(repo="inferrsi", limit=100)
     names = {record["name"] for record in skill_records(ROOT)}
     registry = json.loads((ROOT / "skill-index.json").read_text(encoding="utf-8"))["skills"]
-    assert len(records) == len(names) == 76
+    assert len(records) == len(names) == 77
     assert {record["skill_metadata"]["name"] for record in records} == names
     states = []
     for record in records:
@@ -41,7 +39,7 @@ def test_complete_index_retains_metadata_states_and_unknown_license():
         }
         assert set(record["source_files"]) == set(record["sources"][0]["paths"])
         states.append(metadata.get("status"))
-    assert states.count("candidate") == 71
+    assert states.count("candidate") == 72
     assert states.count(None) == 5
     profiling = next(r for r in records if r["title"] == "ascend-operator-profiling")
     assert profiling["skill_metadata"]["license"] == "UNKNOWN"
@@ -188,7 +186,7 @@ def test_package_index_and_checkout_skill_have_distinct_read_boundaries():
     assert reader.resolve("claim-contract") == "checkout:claim-contract"
     assert reader.read("claim-contract").metadata["kind"] == "skill"
     collected = reader.collect(repo="inferrsi", kind="knowledge", limit=100, material=True)
-    assert len(collected["items"]) == 76
+    assert len(collected["items"]) == 77
     assert not collected["collisions"] and not collected["omissions"]
     assert all(item["authority"] == "package" for item in collected["items"])
     assert all("record.json" in item["resources"] for item in collected["items"])
@@ -218,9 +216,9 @@ def test_index_cli_search_show_context_and_execution_boundary(capsys):
 
 
 def test_hardware_filters_include_only_matching_and_shared_skills():
-    for backend, count in (("ascend", 71), ("cuda", 62), ("cpu", 57)):
+    for backend, count in (("ascend", 72), ("cuda", 63), ("cpu", 58)):
         records = search_knowledge(repo="inferrsi", backend=backend, limit=100)
         assert len(records) == count
         assert all(record["backend"] in (backend, "common") for record in records)
     assert not (PROJECT / "skills").exists()
-    assert len(list(ROOT.rglob("SKILL.md"))) == 120
+    assert len(list(ROOT.rglob("SKILL.md"))) == 124
diff --git a/tests/test_skill_library.py b/tests/test_skill_library.py
index efcca3d54..fc23e36d4 100644
--- a/tests/test_skill_library.py
+++ b/tests/test_skill_library.py
@@ -1,5 +1,3 @@
-"""Portable knowledge integration; these checks do not validate inference claims."""
-
 import json
 import re
 import subprocess
@@ -154,7 +152,7 @@ def test_library_root_exposes_only_owner_categories_and_required_entrypoints():
     assert {path.name for path in LIBRARY.iterdir() if path.is_dir()} == {
         "hardware", "frameworks", "repositories", "methods", "workflows",
     }
-    assert len(list(LIBRARY.rglob("SKILL.md"))) == 120
+    assert len(list(LIBRARY.rglob("SKILL.md"))) == 124
     assert not any(path.name == "zh-CN" for path in LIBRARY.rglob("*") if path.is_dir())
     references = LIBRARY / "workflows/inference-engine-research/references"
     assert (references / "import.md").is_file()