kmshihab7878 Doc Gap Analyzer — Documentation Department Intelligence Engine
Find what's undocumented, what's stale, and what users can't find. UAOP Stage 1 + Stage 5 for Documentation.
When to use
- Before releases — "is everything documented?"
- After new features — "what docs need updating?"
- Quarterly doc health review
- Onboarding review — "can a new person get started?"
Pipeline
Step 1: Coverage Scan
- Compare public API endpoints against doc pages (every endpoint needs a doc)
- Compare Python public functions against docstrings
- Check CLAUDE.md / README age vs last code change
- Verify all skills have SKILL.md files
- Check knowledge graph freshness (memory MCP entities)
Step 2: Staleness Detection
- Docs modified > 90 days ago while code changed → flag as stale
- Code examples that reference deprecated APIs → flag as broken
- Links that return 404 → flag for fix
Step 3: Gap Classification
| Gap Type | Priority |
|---|---|
| Public API with no docs | P0 |
| Stale docs (code changed, docs didn't) | P1 |
| No onboarding guide | P1 |
| Missing code examples | P2 |
| Broken internal links | P2 |
| Knowledge graph gaps (orphan entities) | P3 |
Step 4: Produce Gap Report
- Coverage metrics (% of APIs documented, % of functions with docstrings)
- Top 10 gaps by priority
- Staleness report (docs that need refresh)
- Trend: improving or regressing?
Step 5: Self-Improvement Loop
- Track doc-to-code sync ratio over time
- Measure onboarding time (proxy for doc quality)
- Auto-generated docs that need zero editing → template as standard
Cadence
WEEKLY: Staleness scan + broken link check
MONTHLY: Full coverage report
QUARTERLY: Onboarding review + doc strategy update
Agents
| Agent | Role |
|---|---|
| repo-index (L0) | Scans codebase structure |
| knowledge-graph-guide (L0) | Checks knowledge graph completeness |
| documenter (L6) | Writes docs targeting gaps |
| technical-writer (L6) | Reviews doc quality |