How brefledev/claudemd-prove-it works
claudemd-prove-it is a dependency-free Node.js CLI that audits Claude Code memory files against settings-based enforcement. The main workflow discovers memory and settings, extracts rules and annotations, evaluates enforcement, then reports findings and context costs. The README also documents import, duplication, contradiction, and structure checks, plus an optional judgment pass using PROMPT.md. The source excerpt shows orchestration and parts of discovery and scoring; unsampled capability links are included only where documented.
Main entry points: prove-it.mjs CLI orchestration · PROMPT.md Judgment prompt
Components 17
Project discovery
- CLI orchestration
prove-it.mjs - Settings discovery
prove-it.mjs - Memory discovery
prove-it.mjs - Import resolution
prove-it.mjs
Rule audit
- Rule extraction
prove-it.mjs - Enforcement mapping
prove-it.mjs - Annotation checks
prove-it.mjs - Duplication checks
prove-it.mjs - Structure lint
prove-it.mjs - Scoped-rule checks
prove-it.mjs - Context estimates
prove-it.mjs
Results
- Audit report
prove-it.mjs - Judgment prompt
PROMPT.md
Other components
- Engineer
- Project files
- CLI output
- Claude Code
Connections 28
- Engineer → CLI orchestration: invokes
- CLI orchestration → Settings discovery: discovers
- CLI orchestration → Memory discovery: discovers
- Settings discovery → Project files: reads settings
- Memory discovery → Project files: reads memory
- Memory discovery → Import resolution: resolves imports
- Import resolution → Project files: reads imports
- CLI orchestration → Rule extraction: extracts rules
- CLI orchestration → Enforcement mapping: maps enforcement
- Settings discovery → Enforcement mapping: supplies controls
- Rule extraction → Enforcement mapping: supplies rules
- Rule extraction → Annotation checks: checks claims
- CLI orchestration → Duplication checks: checks rules
- CLI orchestration → Structure lint: checks clarity
- CLI orchestration → Scoped-rule checks: checks scope
- CLI orchestration → Context estimates: estimates cost
- Enforcement mapping → Audit report: supplies verdicts
- Annotation checks → Audit report: supplies results
- Duplication checks → Audit report: supplies findings
- Structure lint → Audit report: supplies findings
- Scoped-rule checks → Audit report: supplies findings
- Context estimates → Audit report: supplies estimates
- CLI orchestration → Audit report: renders results
- Audit report → CLI output: prints
- Engineer → Judgment prompt: provides prompt
- Judgment prompt → Claude Code: guides review
- Claude Code → CLI orchestration: runs audit
- Claude Code → Project files: reads hook scripts