Kanboard Doctor

Diagnose parser issues across all workflow documents. Non-destructive — only suggests fixes, asks user to confirm.

Phase 1: Run All Parsers

Execute each parser against its document type directory:

1. Load the parser registry from @pi-unipi/kanboard
2. Run registry.parseAll(".unipi/docs") to parse all documents
3. Collect all ParsedDoc results including their warnings arrays

Phase 2: Collect Errors

Group warnings and errors by file with line numbers:

1. For each ParsedDoc with warnings.length > 0:
   • Group by filePath
   • Include line numbers where available
   • Categorize: malformed checkboxes, empty fields, parse failures
2. Also flag documents that returned 0 items (may indicate parsing failure)

Phase 3: Present Report

Show a structured error report:

📋 Kanboard Doctor Report

Files scanned: N
Files with issues: M

📄 .unipi/docs/specs/example.md
  ⚠ Line 15: Empty checkbox text
  ⚠ Line 23: Malformed checkbox (missing bracket)

📄 .unipi/docs/plans/old-plan.md
  ⚠ Line 5: Empty task name after status

Phase 4: Fix One by One

For each issue, suggest a fix and ask user to confirm:

1. Show the problematic line with context
2. Suggest the corrected version
3. Ask: "Apply this fix? (y/n)"
4. If yes, apply the fix using the edit tool
5. Move to next issue

Non-destructive rules:

• Never modify without asking
• Show before/after for each change
• Allow skipping individual fixes
• Allow "fix all" for simple patterns (e.g., trailing whitespace)

Phase 5: Re-validate

After each fix, re-run the parser on the modified file:

1. Parse the file again
2. Verify the specific error is resolved
3. Check that no new errors were introduced
4. Report: "✓ Fixed" or "✗ Still has issues"

After all fixes, run full registry.parseAll() to confirm clean state.