From 06c59aa0264d1417d6037a9827a3b417797c9ddf Mon Sep 17 00:00:00 2001 From: Nihit Kasabwala <> Date: Tue, 27 Oct 2026 21:41:31 -0400 Subject: [PATCH 1/4] feat(core-rules): move engineering practices from constraints into core rules The constraints skill reaches an agent only if the agent loads it. context/k-core-rules.md is always loaded by all 11 agents, so it now has an "Engineering practices" section after Writing habits: follow codebase patterns, keep backward compatibility, comment the why, structured logging with correlation IDs when adding or changing logging, research unknown systems first, state documented vs inference vs guess, and confirm tests pass. skills/constraints/SKILL.md points at the core rules for these and keeps maker-checker, the gate tiers, severity ordering, and the circuit breaker. Its description keeps the "MUST USE for every interaction" trigger and now says what the skill still defines. Adds 592 bytes per agent (6512 across 11 agents). Verified: - make build: ok - make synth: ok; section present in all 11 dist/claude/agents/*.md; kiro-cli-v2 and kiro-v3 agents reference file://context/k-core-rules.md and dist context copies match the source - cd cli && TMPDIR=/tmp make test: 1779 passed, 0 failed (first run hit the known resolve_cwd_json_envelope flake; passed 3/3 in isolation and on rerun) - make test-skills-sh: 10 passed - make claude-plugin-check: ok - make test-py (venv, pytest 8.3.4, PyYAML 6.0.2): 222 passed - make kiro-power-check (same venv): all passed --- CHANGELOG.md | 2 ++ context/k-core-rules.md | 10 ++++++++++ skills/constraints/SKILL.md | 20 ++++++++------------ 3 files changed, 20 insertions(+), 12 deletions(-) diff --git a/CHANGELOG.md b/CHANGELOG.md index 01b8a6a0..db6163e5 100644 --- a/CHANGELOG.md +++ b/CHANGELOG.md @@ -10,6 +10,8 @@ not per individual commit. ### Changed +- The always-loaded core rules now carry the engineering practices from the `constraints` skill, + so agents get them without loading the skill. - The orchestrators (`konductor`, `konductor-mux-orchestrator`, `konductor-cmux-orchestrator`) may now read files, search the codebase, run read-only shell commands, and answer lookup questions themselves instead of spawning a specialist for each. One table in diff --git a/context/k-core-rules.md b/context/k-core-rules.md index f0446b71..57448fc8 100644 --- a/context/k-core-rules.md +++ b/context/k-core-rules.md @@ -18,3 +18,13 @@ These rules apply to every Konductor agent in every interaction. Violating a NEV ## Writing habits In sentences you write or rewrite, do not use em or en dashes, AI vocabulary, or filler, even when nearby text does (see the `humanize-writing` skill for how to rewrite them). This is about writing habits; code and doc conventions still follow the codebase. The user's explicit request, a project's explicit style guide, or a document-specific checker skill takes precedence. Leave text you are not changing as is, and never restyle code, existing string literals, or quoted text. + +## Engineering practices + +- Follow the patterns and conventions already in the codebase. +- Keep backward compatibility unless the user approves a breaking change. +- Comment non-obvious logic with the why, not the what. +- When you add or change logging, make it structured and include correlation IDs. +- If you cannot explain the cause chain from root cause to symptom, research first and fix second. +- Say whether a claim rests on documented knowledge (you read the code or docs), inference, or a guess. +- Before calling a task done, confirm the tests pass, not only that the code compiles. diff --git a/skills/constraints/SKILL.md b/skills/constraints/SKILL.md index ae5a2da5..454517a5 100644 --- a/skills/constraints/SKILL.md +++ b/skills/constraints/SKILL.md @@ -1,6 +1,6 @@ --- name: constraints -description: MUST USE for every interaction across all Konductor agents. Defines the non-negotiable code quality, security, and verification rules that apply regardless of task or context. +description: MUST USE for every interaction across all Konductor agents. Defines the maker-checker, quality gate, severity ordering, and failure recovery rules; the always-loaded core rules hold the rest. version: 1.0.0 tags: [skill, behavioral, constraints, quality, security] --- @@ -24,12 +24,12 @@ The NEVER DO rules live in the core rules, which every agent loads at session st ### ALWAYS DO 1. **Read before edit** — Always read the complete file before making changes -2. **Follow existing patterns** — Always match the conventions already in the codebase. Writing habits are covered in `k-core-rules`. +2. **Follow existing patterns**: see Engineering practices and Writing habits in the core rules 3. **Maker-checker** — Always run the maker-checker pattern: generate artifact → validate with the corresponding checker skill before presenting to the user -4. **Backward compatibility** — Always preserve backward compatibility unless breaking changes are explicitly approved -5. **Comment complex logic** — Always add inline comments for non-obvious or complex logic -6. **Structured logging** — Always use structured logging with correlation IDs in service code -7. **Verify before done** — Always verify changes compile and tests pass before declaring a task complete +4. **Backward compatibility**: see Engineering practices in the core rules +5. **Comment complex logic**: see Engineering practices in the core rules +6. **Structured logging**: see Engineering practices in the core rules +7. **Verify before done**: see NEVER DO 3 and 5 and Engineering practices in the core rules 8. **Severity ordering** — Always present findings as CRITICAL → IMPORTANT → SUGGESTION ## Failure Recovery @@ -44,15 +44,11 @@ After 2 failed fix attempts for the same error, **STOP**. Do not attempt a third ### Unknown Systems -If you cannot explain the cause chain from root cause → symptom, **research FIRST, fix SECOND**. Do not guess at fixes for systems you don't understand. +See Engineering practices in the core rules: research first, fix second. ### Acknowledge Uncertainty -When responding, state whether you are working from: - -- **Documented knowledge** — you read the relevant code/docs -- **Inference** — you are reasoning from patterns but haven't verified -- **Guess** — you are speculating and the user should verify +See Engineering practices in the core rules: say whether you are working from documented knowledge, inference, or a guess. ## Quality Gate From 19e88563144ffe49bc3e85e546435d452d66924e Mon Sep 17 00:00:00 2001 From: Nihit <24500178+knihit@users.noreply.github.com> Date: Wed, 28 Oct 2026 03:23:58 +0000 Subject: [PATCH 2/4] fix(core-rules): scope the structured logging rule to service code The promoted rule applied to every logging change, including scripts, CLIs and tests. Restore the service-code scope from the constraints skill and tell agents to follow the existing logging style elsewhere. --- context/k-core-rules.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/context/k-core-rules.md b/context/k-core-rules.md index 57448fc8..8e268a88 100644 --- a/context/k-core-rules.md +++ b/context/k-core-rules.md @@ -24,7 +24,7 @@ In sentences you write or rewrite, do not use em or en dashes, AI vocabulary, or - Follow the patterns and conventions already in the codebase. - Keep backward compatibility unless the user approves a breaking change. - Comment non-obvious logic with the why, not the what. -- When you add or change logging, make it structured and include correlation IDs. +- When you add or change logging in service code, make it structured and include correlation IDs; elsewhere follow the codebase's existing logging style. - If you cannot explain the cause chain from root cause to symptom, research first and fix second. - Say whether a claim rests on documented knowledge (you read the code or docs), inference, or a guess. - Before calling a task done, confirm the tests pass, not only that the code compiles. From e5416ada179d2c08233dc0355a5b392b4fa46c26 Mon Sep 17 00:00:00 2001 From: Nihit <24500178+knihit@users.noreply.github.com> Date: Wed, 28 Oct 2026 03:23:58 +0000 Subject: [PATCH 3/4] docs(user-guide): describe engineering practices in the core rules The guide still said the core rules hold only the NEVER DO rules and writing habits. Name the Engineering practices section too, and describe the constraints skill by what it keeps: read before edit, maker-checker, severity ordering, the circuit breaker and the quality gate. Update the matching sync-html-content.py entries so their old text is what the bundles carry today, then regenerate both bundles and the manifest. --- docs/index.html | 2 +- docs/site/.user-guide.manifest | 8 ++++---- docs/site/user-guide.html | 2 +- docs/user-guide/appendix/contributing.md | 2 +- docs/user-guide/concepts.md | 4 ++-- docs/user-guide/faq.md | 2 +- docs/user-guide/skills.md | 2 +- tools/build-user-guide/sync-html-content.py | 16 ++++++++-------- 8 files changed, 19 insertions(+), 19 deletions(-) diff --git a/docs/index.html b/docs/index.html index 9b2bcffb..ea9e74db 100644 --- a/docs/index.html +++ b/docs/index.html @@ -381,7 +381,7 @@ \ No newline at end of file diff --git a/docs/site/.user-guide.manifest b/docs/site/.user-guide.manifest index 34ac1d8c..8194e80b 100644 --- a/docs/site/.user-guide.manifest +++ b/docs/site/.user-guide.manifest @@ -1,6 +1,6 @@ # Generated by tools/build-user-guide/build-user-guide.sh — do not edit. -content_sha256=1f27c530f1d309023e979c2098f6050628e99f94fcc0c22fd0d54924e44bf32d -built_at=2026-09-23T15:08:36Z +content_sha256=0e76b4265ae1e4109416cea8fa10fa699324a3c69c0f7ab52be36f27dd1ae500 +built_at=2026-10-28T03:20:57Z runtime=deterministic -source_commit=6a5b6f0e93330f0baffe85427f81edca8d0d677b -output_sha256=9aae3ee3ccf9fa2b46f3afee49bf3f45d303030df19e0010af68177d79e9076c +source_commit=06c59aa0264d1417d6037a9827a3b417797c9ddf +output_sha256=1c6b1437f7230c5b53ba5fd877ec2c5c3c97709f2de123dec9b87630ec7ac969 diff --git a/docs/site/user-guide.html b/docs/site/user-guide.html index 9b2bcffb..ea9e74db 100644 --- a/docs/site/user-guide.html +++ b/docs/site/user-guide.html @@ -381,7 +381,7 @@ \ No newline at end of file diff --git a/docs/user-guide/appendix/contributing.md b/docs/user-guide/appendix/contributing.md index 4ce41b23..9828584a 100644 --- a/docs/user-guide/appendix/contributing.md +++ b/docs/user-guide/appendix/contributing.md @@ -30,7 +30,7 @@ Five files control five different things. Pick the smallest one that achieves wh | Change | Edit | Effect | | --- | --- | --- | -| Rules every agent must follow | `context/k-core-rules.md` | NEVER DO rules and writing habits, loaded at session start for every agent | +| Rules every agent must follow | `context/k-core-rules.md` | NEVER DO rules, writing habits, and engineering practices, loaded at session start for every agent | | Which specialist handles which kind of request | `context/k-orchestrator-routing-rules.md` | Routing, loaded at session start for all three orchestrators | | The steps of one workflow | `agent-sops/.sop.md` | That SOP's steps, constraints, and quality gates | | How an agent reasons about one topic | `skills//SKILL.md` | Guidance that agent loads on demand | diff --git a/docs/user-guide/concepts.md b/docs/user-guide/concepts.md index dc8bb073..8f327595 100644 --- a/docs/user-guide/concepts.md +++ b/docs/user-guide/concepts.md @@ -153,8 +153,8 @@ A **context file** is Markdown loaded into an agent's memory *at session start* active for the whole session — unlike a skill, which loads on demand. Context files are for rules that must never be missed. -There are two today. `context/k-core-rules.md` holds the NEVER DO rules and the writing-habits -rule, and all 11 agents load it. `context/k-orchestrator-routing-rules.md` is loaded by all three +There are two today. `context/k-core-rules.md` holds the NEVER DO rules, the writing-habits +rule, and the engineering practices, and all 11 agents load it. `context/k-orchestrator-routing-rules.md` is loaded by all three orchestrators. The routing rules file contains the capability-to-agent routing table and three `[CODE RED]` rules, including this one: diff --git a/docs/user-guide/faq.md b/docs/user-guide/faq.md index db69326c..f2bb322d 100644 --- a/docs/user-guide/faq.md +++ b/docs/user-guide/faq.md @@ -134,7 +134,7 @@ Common edits, in ascending order of blast radius: | Edit | File | Effect | | --- | --- | --- | -| Change the core rules | `context/k-core-rules.md` | NEVER DO rules and writing habits, loaded at session start for every agent | +| Change the core rules | `context/k-core-rules.md` | NEVER DO rules, writing habits, and engineering practices, loaded at session start for every agent | | Change routing | `context/k-orchestrator-routing-rules.md` | Which specialist gets which kind of request | | Change a procedure | `agent-sops/.sop.md` | The steps, constraints, and gates of one workflow | | Change knowledge | `skills//SKILL.md` | How an agent reasons about one topic | diff --git a/docs/user-guide/skills.md b/docs/user-guide/skills.md index cbc093d7..c7e2a574 100644 --- a/docs/user-guide/skills.md +++ b/docs/user-guide/skills.md @@ -42,7 +42,7 @@ Loaded to shape *how* an agent works rather than what it produces. Several are o | Skill | What it does | Reach for it when | Declared by | | --- | --- | --- | --- | -| `constraints` | Hard behavioural rules for every agent: what they must always do, across code quality, security, and verification. The NEVER DO rules live in `context/k-core-rules.md`. | You never invoke it. Claude Code preloads it only into subagents; otherwise it loads on demand. | all 11 | +| `constraints` | Workflow rules for every agent: read before edit, maker-checker validation, severity ordering, the 2-strike circuit breaker, and the quality gate. The NEVER DO rules, writing habits, and engineering practices live in `context/k-core-rules.md`. | You never invoke it. Claude Code preloads it only into subagents; otherwise it loads on demand. | all 11 | | `about-konductor` | Gets a new user from nothing to a working session: the real `konductor` CLI command surface, example orchestrator prompts, and the agent/skill/SOP model. | Someone asks what this is, how to install it, or what they can ask for. Backs the SOP of the same name. | all 3 orchestrators | | `sop-state-management` | Session and checklist mechanics for long-running SOPs: minting or resuming a session by project identity, per-phase status, and serialized fix-cycle counters. | You are running or resuming `k-full-sdlc`. It owns the state file schema and the resume-discovery walk. | all 3 orchestrators | | `delegation-protocol` | The authoritative delegation format: the 8-field prompt (target agent + 7 sections), the agent registry, parallel-execution rules, and handoff patterns. | Read it to understand what the orchestrator sends a specialist. Edit it to change routing — it outranks the routing-rules context file. | all 3 orchestrators | diff --git a/tools/build-user-guide/sync-html-content.py b/tools/build-user-guide/sync-html-content.py index 4a889670..586e02c6 100755 --- a/tools/build-user-guide/sync-html-content.py +++ b/tools/build-user-guide/sync-html-content.py @@ -80,8 +80,8 @@ 'search/read, region list, and regional availability.'), # ---- every agent loads context/k-core-rules.md --------------------------- ("concepts context file paragraph", - "There is one today: context/k-orchestrator-routing-rules.md, loaded by all three orchestrators. It contains", - "There are two today. context/k-core-rules.md holds the NEVER DO rules and the writing-habits rule, and all 11 agents load it. context/k-orchestrator-routing-rules.md is loaded by all three orchestrators. The routing rules file contains"), + "There are two today. context/k-core-rules.md holds the NEVER DO rules and the writing-habits rule, and all 11 agents load it. context/k-orchestrator-routing-rules.md is loaded by all three orchestrators. The routing rules file contains", + "There are two today. context/k-core-rules.md holds the NEVER DO rules, the writing-habits rule, and the engineering practices, and all 11 agents load it. context/k-orchestrator-routing-rules.md is loaded by all three orchestrators. The routing rules file contains"), ("concepts context files row", '{ k: "Context files — 1", v: "The orchestrator routing rules, always active" }', '{ k: "Context files — 2", v: "The core rules for every agent and the orchestrator routing rules, always active" }'), @@ -100,14 +100,14 @@ ("troubleshooting context count", "it should report 1 registered.", "it should report 2 registered."), ("contributing: five files", "Four files control four different things.", "Five files control five different things."), ("contributing: core rules row", - 'h1: "EDIT", h2: "EFFECT", rows: [\n { k: "context/k-orchestrator-routing-rules.md", v: "Which specialist', - 'h1: "EDIT", h2: "EFFECT", rows: [\n { k: "context/k-core-rules.md", v: "Rules every agent must follow, NEVER DO rules and writing habits, loaded at session start for every agent" },\n { k: "context/k-orchestrator-routing-rules.md", v: "Which specialist'), + '{ k: "context/k-core-rules.md", v: "Rules every agent must follow, NEVER DO rules and writing habits, loaded at session start for every agent" }', + '{ k: "context/k-core-rules.md", v: "Rules every agent must follow, NEVER DO rules, writing habits, and engineering practices, loaded at session start for every agent" }'), ("faq: core rules row", - 'rows: [\n { k: "context/k-orchestrator-routing-rules.md", v: "Change routing', - 'rows: [\n { k: "context/k-core-rules.md", v: "Change the core rules, the NEVER DO rules and writing habits loaded at session start for every agent" },\n { k: "context/k-orchestrator-routing-rules.md", v: "Change routing'), + '{ k: "context/k-core-rules.md", v: "Change the core rules, the NEVER DO rules and writing habits loaded at session start for every agent" }', + '{ k: "context/k-core-rules.md", v: "Change the core rules, the NEVER DO rules, writing habits, and engineering practices loaded at session start for every agent" }'), ("skills catalog: constraints row", - "every agent \u2014 what they must never and always do, across code quality, security, and verification.", - "every agent: what they must always do, across code quality, security, and verification. The NEVER DO rules live in context/k-core-rules.md."), + "Hard behavioural rules for every agent: what they must always do, across code quality, security, and verification. The NEVER DO rules live in context/k-core-rules.md.", + "Workflow rules for every agent: read before edit, maker-checker validation, severity ordering, the 2-strike circuit breaker, and the quality gate. The NEVER DO rules, writing habits, and engineering practices live in context/k-core-rules.md."), ("skills catalog: constraints reach", "REACH FOR IT: Always on. You never invoke it; it is the floor every agent stands on.", "REACH FOR IT: You never invoke it. Claude Code preloads it only into subagents; otherwise it loads on demand."), From e778319fd6056ce92044a4ea35c0b59c0d6a2615 Mon Sep 17 00:00:00 2001 From: Nihit <24500178+knihit@users.noreply.github.com> Date: Wed, 28 Oct 2026 03:51:13 +0000 Subject: [PATCH 4/4] fix(core-rules): scope the test and uncertainty rules, and move read-before-edit Every agent loads the core rules, orchestrators and non-code agents included. The test rule now covers code changes, and an orchestrator gets the evidence from the specialist instead of running tests. The uncertainty rule applies to unverified claims that matter to a decision. Read-before-edit moves from the constraints skill, and constraints ALWAYS 7 names the NEVER DO rules it points at instead of their numbers. --- context/k-core-rules.md | 5 +++-- docs/index.html | 2 +- docs/site/.user-guide.manifest | 8 ++++---- docs/site/user-guide.html | 2 +- docs/user-guide/skills.md | 2 +- skills/constraints/SKILL.md | 6 +++--- tools/build-user-guide/sync-html-content.py | 4 ++-- 7 files changed, 15 insertions(+), 14 deletions(-) diff --git a/context/k-core-rules.md b/context/k-core-rules.md index 8e268a88..c4dfea2f 100644 --- a/context/k-core-rules.md +++ b/context/k-core-rules.md @@ -21,10 +21,11 @@ In sentences you write or rewrite, do not use em or en dashes, AI vocabulary, or ## Engineering practices +- Read the whole file before you edit it. - Follow the patterns and conventions already in the codebase. - Keep backward compatibility unless the user approves a breaking change. - Comment non-obvious logic with the why, not the what. - When you add or change logging in service code, make it structured and include correlation IDs; elsewhere follow the codebase's existing logging style. - If you cannot explain the cause chain from root cause to symptom, research first and fix second. -- Say whether a claim rests on documented knowledge (you read the code or docs), inference, or a guess. -- Before calling a task done, confirm the tests pass, not only that the code compiles. +- When a claim matters to a decision and you have not verified it, say whether it rests on documented knowledge (you read the code or docs), inference, or a guess. +- Before calling a code change done, confirm the tests pass, not only that the code compiles. An orchestrator gets that evidence from the specialist rather than running the tests. diff --git a/docs/index.html b/docs/index.html index ea9e74db..5372bda5 100644 --- a/docs/index.html +++ b/docs/index.html @@ -381,7 +381,7 @@ \ No newline at end of file diff --git a/docs/site/.user-guide.manifest b/docs/site/.user-guide.manifest index 8194e80b..5e95161b 100644 --- a/docs/site/.user-guide.manifest +++ b/docs/site/.user-guide.manifest @@ -1,6 +1,6 @@ # Generated by tools/build-user-guide/build-user-guide.sh — do not edit. -content_sha256=0e76b4265ae1e4109416cea8fa10fa699324a3c69c0f7ab52be36f27dd1ae500 -built_at=2026-10-28T03:20:57Z +content_sha256=a1a5035e77885341a6d15f4f65834a562975948b00a12be53195ac2b2381a905 +built_at=2026-10-28T03:50:11Z runtime=deterministic -source_commit=06c59aa0264d1417d6037a9827a3b417797c9ddf -output_sha256=1c6b1437f7230c5b53ba5fd877ec2c5c3c97709f2de123dec9b87630ec7ac969 +source_commit=e5416ada179d2c08233dc0355a5b392b4fa46c26 +output_sha256=b3da85ae15fdba256f6a2d5bd6a478404f9e0ec0706f933a42b75b2907399427 diff --git a/docs/site/user-guide.html b/docs/site/user-guide.html index ea9e74db..5372bda5 100644 --- a/docs/site/user-guide.html +++ b/docs/site/user-guide.html @@ -381,7 +381,7 @@ \ No newline at end of file diff --git a/docs/user-guide/skills.md b/docs/user-guide/skills.md index c7e2a574..23a5b7f4 100644 --- a/docs/user-guide/skills.md +++ b/docs/user-guide/skills.md @@ -42,7 +42,7 @@ Loaded to shape *how* an agent works rather than what it produces. Several are o | Skill | What it does | Reach for it when | Declared by | | --- | --- | --- | --- | -| `constraints` | Workflow rules for every agent: read before edit, maker-checker validation, severity ordering, the 2-strike circuit breaker, and the quality gate. The NEVER DO rules, writing habits, and engineering practices live in `context/k-core-rules.md`. | You never invoke it. Claude Code preloads it only into subagents; otherwise it loads on demand. | all 11 | +| `constraints` | Workflow rules for every agent: maker-checker validation, severity ordering, the 2-strike circuit breaker, and the quality gate. The NEVER DO rules, writing habits, and engineering practices live in `context/k-core-rules.md`. | You never invoke it. Claude Code preloads it only into subagents; otherwise it loads on demand. | all 11 | | `about-konductor` | Gets a new user from nothing to a working session: the real `konductor` CLI command surface, example orchestrator prompts, and the agent/skill/SOP model. | Someone asks what this is, how to install it, or what they can ask for. Backs the SOP of the same name. | all 3 orchestrators | | `sop-state-management` | Session and checklist mechanics for long-running SOPs: minting or resuming a session by project identity, per-phase status, and serialized fix-cycle counters. | You are running or resuming `k-full-sdlc`. It owns the state file schema and the resume-discovery walk. | all 3 orchestrators | | `delegation-protocol` | The authoritative delegation format: the 8-field prompt (target agent + 7 sections), the agent registry, parallel-execution rules, and handoff patterns. | Read it to understand what the orchestrator sends a specialist. Edit it to change routing — it outranks the routing-rules context file. | all 3 orchestrators | diff --git a/skills/constraints/SKILL.md b/skills/constraints/SKILL.md index 454517a5..7ffb38e9 100644 --- a/skills/constraints/SKILL.md +++ b/skills/constraints/SKILL.md @@ -23,13 +23,13 @@ The NEVER DO rules live in the core rules, which every agent loads at session st ### ALWAYS DO -1. **Read before edit** — Always read the complete file before making changes +1. **Read before edit**: see Engineering practices in the core rules 2. **Follow existing patterns**: see Engineering practices and Writing habits in the core rules 3. **Maker-checker** — Always run the maker-checker pattern: generate artifact → validate with the corresponding checker skill before presenting to the user 4. **Backward compatibility**: see Engineering practices in the core rules 5. **Comment complex logic**: see Engineering practices in the core rules 6. **Structured logging**: see Engineering practices in the core rules -7. **Verify before done**: see NEVER DO 3 and 5 and Engineering practices in the core rules +7. **Verify before done**: see "No broken commits" and "No broken state" under NEVER DO, and Engineering practices, in the core rules 8. **Severity ordering** — Always present findings as CRITICAL → IMPORTANT → SUGGESTION ## Failure Recovery @@ -48,7 +48,7 @@ See Engineering practices in the core rules: research first, fix second. ### Acknowledge Uncertainty -See Engineering practices in the core rules: say whether you are working from documented knowledge, inference, or a guess. +See Engineering practices in the core rules: for an unverified claim that matters, say whether it rests on documented knowledge, inference, or a guess. ## Quality Gate diff --git a/tools/build-user-guide/sync-html-content.py b/tools/build-user-guide/sync-html-content.py index 586e02c6..76c2dfa8 100755 --- a/tools/build-user-guide/sync-html-content.py +++ b/tools/build-user-guide/sync-html-content.py @@ -106,8 +106,8 @@ '{ k: "context/k-core-rules.md", v: "Change the core rules, the NEVER DO rules and writing habits loaded at session start for every agent" }', '{ k: "context/k-core-rules.md", v: "Change the core rules, the NEVER DO rules, writing habits, and engineering practices loaded at session start for every agent" }'), ("skills catalog: constraints row", - "Hard behavioural rules for every agent: what they must always do, across code quality, security, and verification. The NEVER DO rules live in context/k-core-rules.md.", - "Workflow rules for every agent: read before edit, maker-checker validation, severity ordering, the 2-strike circuit breaker, and the quality gate. The NEVER DO rules, writing habits, and engineering practices live in context/k-core-rules.md."), + "Workflow rules for every agent: read before edit, maker-checker validation, severity ordering, the 2-strike circuit breaker, and the quality gate. The NEVER DO rules, writing habits, and engineering practices live in context/k-core-rules.md.", + "Workflow rules for every agent: maker-checker validation, severity ordering, the 2-strike circuit breaker, and the quality gate. The NEVER DO rules, writing habits, and engineering practices live in context/k-core-rules.md."), ("skills catalog: constraints reach", "REACH FOR IT: Always on. You never invoke it; it is the floor every agent stands on.", "REACH FOR IT: You never invoke it. Claude Code preloads it only into subagents; otherwise it loads on demand."),