review-context

Overview: Comprehensive review of a project's codified context infrastructure. Evaluates completeness, quality, and coverage to identify gaps and improvements.

Workflow:

1. Infrastructure Inventory

   • Check which context files exist in .claude/context/
   • Compute completeness score (constitution, trigger tables, failure modes, subsystem map, specs)
   • List any missing core files
   • Count spec files and their sizes

2. Coverage Analysis

   • Compare subsystem-map.md against actual directory structure — which subsystems lack documentation?
   • Check trigger-tables.md coverage — what % of source directories have routing rules?
   • Evaluate failure-modes.md entries — how many categories have entries?
   • Cross-reference specs/ with subsystem map — which subsystems lack specs?

3. Quality Assessment Using criteria from ${CLAUDE_PLUGIN_ROOT}/skills/review-context/references/coverage-criteria.md:

   • Constitution: All 10 sections populated? Commands verified? Stack current?
   • Trigger tables: Routes specific enough? Rationale provided?
   • Failure modes: Entries have all 4 fields (symptom, cause, fix, prevention)?
   • Subsystem map: Boundaries clear? Dependencies documented?

4. Growth Metrics (if git history available)

   • Context files creation date
   • Number of updates since creation
   • Ratio of codebase changes to context updates
   • Session count since last audit

5. Recommendations Report Generate prioritized recommendations:

   • Critical: Missing core files, broken references
   • High: Undocumented subsystems with recent activity
   • Medium: Stale specs, incomplete failure modes
   • Low: Cosmetic improvements, optional enrichments

6. Output Format Present as a structured health report with scores and actionable items.