This repository was archived by the owner on Jul 15, 2026. It is now read-only.
Conversation
session_start/end were SKILL.md instructions that LLM could skip (especially after compaction), causing "no such file" errors when session_end ran without prior session_start. Now a PreToolUse/PostToolUse hook handles both automatically for all x-kit skills, with mkdir -p guaranteeing the traces directory exists. Semantic entries (agent_step) remain LLM best-effort via SKILL.md.
x-kit init copies hooks and merges settings.json from marketplace. x-kit update now reminds users to run init for hook updates.
- x-build: Cost Engine v2 — adaptive model routing with outcome feedback loop - cost-engine.mjs extracted from core.mjs (modular architecture) - cost-learner.mjs for adaptive learning (90-day window, success-rate scoring) - model_learned config key with 4-level priority chain - budget: rolling window, cached spend, per-project limits - forecast: token-actuals.json for data-driven estimates - escalation cascade with quality threshold gates - appendMetric spinlock for concurrent write safety - routing correlation IDs (ce-XXXXXXXX) - MODEL_COSTS updated for Claude 4.x pricing - x-kit: marketplace sync (cost-engine, cost-learner, core, tasks)
…ckBudget, static import - Extract root.mjs with ROOT resolution logic (eliminates duplication) - Extract config-loader.mjs with loadSharedConfig (eliminates duplication) - Rewire cost-engine.mjs to use shared root + config modules - Split checkBudget (130 LOC) into scanMetrics + evaluateBudget + mergeProjectBudget - Replace dynamic import of cost-learner.mjs with static import - refreshModelLearned converted from async to sync - 542 tests pass, zero regressions
- x-build: internal refactoring — extract root.mjs + config-loader.mjs, split checkBudget, static import for cost-learner, evaluateEscalation unknown model guard, mergeProjectBudget dedup
- Gantt Timeline: SVG-based model swim lanes with parallel execution view
- Cost Waterfall: Chart.js stepped line showing cumulative cost over time
- Model Distribution: doughnut chart with calls/cost/duration per model
- Chart.js CDN added, dark theme matched to BRUTALIST EDITION
- Empty data graceful handling ("No agent data for charts")
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
session_start/end were SKILL.md instructions that LLM could skip
(especially after compaction), causing "no such file" errors when
session_end ran without prior session_start.
Now a PreToolUse/PostToolUse hook handles both automatically for all
x-kit skills, with mkdir -p guaranteeing the traces directory exists.
Semantic entries (agent_step) remain LLM best-effort via SKILL.md.