Validate Docs

validate-docs -- Code Documentation Quality Validation

Corresponding rule: docs/specs/documentation.md

Input -- Dynamic Target Resolution

Resolve the target directory in the following order:

1. CLAUDE.md "Project Structure" section -- extract source code directories
2. If CLAUDE.md is absent or has no structure: Glob source files (**/*.{ts,tsx,js,jsx,py,java,go,rs}) -- use the top-level directory containing matches
3. If no source files exist: report status as SHALLOW with reason

Default exclusions: node_modules/, .git/, dist/, build/, coverage/ Skill-specific exclusions: *.generated.*, *.g.dart, *.freezed.dart (auto-generated code)

Mandatory output: Target resolution result

Output the following before proceeding to validation. Do not proceed until resolved.

Execution logic

1. Validate TODO/FIXME formatting

   • Grep for the TODO or FIXME keywords
   • Valid format: TODO(@assignee, YYYY-MM): description or FIXME(@assignee, YYYY-MM): description
   • Warn when the (@ + , + 4-digit year pattern is absent
   • TODOs/FIXMEs without an assignee/deadline cannot be tracked and risk permanent neglect

2. Validate HACK removal conditions

   • Grep for the HACK keyword
   • Warn when no removal condition or date (YYYY, vN, #issue-number) is present
   • A HACK without a removal condition causes temporary code to become permanent

3. Prefix usage report

   • Prefix list: WHY, CONSTRAINT, DEPENDS, SECURITY, PERF, MIGRATION
   • Grep the usage count of each prefix
   • Generate a prefix usage ratio report (informational only, no PASS/FAIL verdict)

4. Detect self-evident comments (low confidence)

   • Detection patterns: // declare variable, // increment i, // call function, // return, // loop, // declare
   • Patterns where the comment restates what the code already expresses
   • High false-positive rate, so marked as "suspected items" only (not FAIL)

Mandatory output: Findings verification matrix

Output the following matrix before generating the final report. Do not proceed until every category has been checked.

Status values: PASS (verified clean), NOT_APPLICABLE (no source files found), SKIPPED (plugin issue), SHALLOW (target resolution failed)

Pre-output checklist (required before final output)

Verify every item before drafting the report. If any item is unchecked, go back and complete it.

• Every validation category has a Status value assigned (no empty Status cells)
• Every category whose Status != SKIPPED has an Evidence value
• Every WARN item includes a file path and line number
• Report language matches the user's conversation language

Schema compliance check (required before persisting to .ww-w-ai/standards/)

Verify the JSON output before writing to .ww-w-ai/standards/validate-docs/:

• All "required" fields in schema.json are present and non-empty
• findings[] array includes every detected item
• summary aggregates match the actual number of findings

Output

Produce the validation report in the user's conversation language.

Output the validation results in the following format:

Documentation Quality Report
==================================================

### TODO/FIXME format

| File:line | Type | Description |
|-----------|:----:|-------------|
| src/auth.ts:42 | WARN | TODO without assignee/deadline: `// TODO: fix later` |
| src/order.ts:15 | PASS | `// TODO(@kim, 2026-06): implement pagination` |
| src/payment.ts:88 | WARN | FIXME without assignee: `// FIXME: intermittent timeout` |

### HACK removal conditions

| File:line | Type | Description |
|-----------|:----:|-------------|
| src/legacy.ts:23 | WARN | HACK without removal condition: `// HACK: don't touch` |

### Prefix usage summary

| Prefix | Count |
|--------|:-----:|
| WHY | 5 |
| CONSTRAINT | 3 |
| DEPENDS | 2 |
| SECURITY | 1 |
| PERF | 0 |
| MIGRATION | 0 |

### Suspected items (self-evident comments)

| File:line | Content |
|-----------|---------|
| src/utils.ts:10 | `// declare the variable` |
| src/calc.ts:25 | `// increment i by 1` |

> Suspected items have a high false-positive rate. Check the context to decide whether the comment is truly unnecessary.

--------------------------------------------------
Total: 3 warnings | 2 suspected items

Output persistence

After generating the documentation quality report, persist results to .ww-w-ai/standards/validate-docs/:

1. Create the .ww-w-ai/standards/validate-docs/ directory if it does not exist
2. Write latest.json -- structured result following templates/schema.json
3. Write latest.md -- human-readable report following templates/report.template.md
4. Archive to history/ -- copy latest.json to .ww-w-ai/standards/validate-docs/history/{timestamp}.json

latest.md is produced in the user's conversation language. JSON field names remain in English regardless of language. The JSON output enables machine-parseable history tracking and cross-run comparison. The history/ directory preserves prior runs for trend analysis.

Permission rationale

• Write: restricted to .ww-w-ai/standards/ output persistence. No modification of project source.
• Bash: restricted to read-only git/system queries. No file modification.

Notes

• This skill is read-only. It does not modify or add comments.
• Self-evident comment detection is based on string pattern matching and has a high false-positive rate. Results are marked as "suspected items" only and the final judgment is the user's.
• Auto-generated code (*.generated.*, etc.) is excluded from the search scope.
• The prefix usage summary is informational; a low usage ratio does not lead to a FAIL verdict.
• English-only detection patterns are used. In non-English projects, detection accuracy (especially for self-evident comment detection) may vary.

Spec references

Detailed verification criteria, evidence tables, examples:

• Corresponding rule spec: ../../docs/specs/documentation.md
• Evidence index: ../../docs/evidence/evidence-registry.md