Todo

Todo Management Skill

This skill consolidates all TODOs from two sources:

1. Code comments - TODO: and CLAUDE: markers in src/, tests/, docs/
2. Todo files - Files with "todo" in the filename

Quick Start - Run Scan Script First

ALWAYS run this script first to save tokens:

python .claude/skills/todo/todo_scan.py

This outputs a complete report showing:

• Tracked vs untracked code TODOs
• Todo files and their status
• GitHub Issues with todo label

Quick Report (Default Action)

When user runs /todo, provide a quick summary:

1. Run the scan script
2. Query GitHub Issues: gh issue list --label todo --state open --json number,title,labels
3. Present combined report

Output format:

=== TODO Quick Report ===

Open Issues: X (Y high, Z medium, W low)
In Progress: N
Untracked in code: M

| #   | Pri  | Category | Title                    | Status      |
|-----|------|----------|--------------------------|-------------|
| #45 | HIGH | bug      | GUI Animation Bug        | in-progress |
| #32 | MED  | arch     | Circular imports         | analyzed    |
...

Untracked Code TODOs: M
  src/file.py:123 - TODO: description
  ...

Run `/todo scan` for full scan
Run `/todo track` to create issues for untracked items

Commands

/todo help

Show quick reference of available commands:

/todo              Quick report (open issues, priorities, in-progress)
/todo scan         Full codebase scan for TODOs
/todo sync         Find inconsistencies between code and GitHub
/todo search       Search issues by keyword or label
/todo track        Create GitHub Issues for untracked TODOs
/todo analyze #id  Analyze TODO and update its GitHub Issue
/todo start #id    Mark issue as in-progress
/todo done #id     Close the issue
/todo reject #id   Close with wontfix label

/todo or /todo report

Quick report from script output and GitHub Issues.

/todo scan

Full scan with detailed output:

python .claude/skills/todo/todo_scan.py

/todo sync

Find inconsistencies between code/files and GitHub Issues:

python .claude/skills/todo/todo_scan.py --sync

Checks performed:

Code TODOs:

1. Missing in GitHub - Code TODO has [#X] but issue doesn't exist or lacks todo label
2. Missing in Code - GitHub Issue with todo:code label has no matching code TODO
3. Stale TODOs - Code TODO references a CLOSED issue (should be removed)

File Entries: 4. Missing in GitHub - File entry references issue that doesn't exist or lacks todo label 5. Missing in File - GitHub Issue with todo:file label has no matching file entry 6. Stale Entries - File entry references a CLOSED issue (should be removed) 7. Status Mismatch - File says in-progress but GitHub lacks label, or vice versa

Output format:

=== SYNC REPORT ===

Code: Missing in GitHub (1):
  src/file.py:42 - [#999] issue does not exist

Code: Stale TODOs (1):
  src/old.py:20 - [#30] issue is CLOSED - remove this TODO

File: Status Mismatch (2):
  todo/todo_open.md:5 - [#16] file says 'investigating' but GitHub has 'in-progress' label
  todo/todo_open.md:15 - [#23] file says 'in-progress' but GitHub lacks label

Synced: 13 code TODOs, 25 file entries

Interactive Resolution:

When inconsistencies are found, ask the user how to resolve each one:

1. Status Mismatch - Ask: "Update file to match GitHub, or update GitHub to match file?"

   • Option A: Update file entry status to match GitHub label
   • Option B: Add/remove in-progress label on GitHub to match file
   • Option C: Skip (do nothing)

2. Stale TODOs/Entries - Ask: "Issue is closed. Remove the TODO/entry?"

   • Option A: Remove the code comment or file entry
   • Option B: Skip (keep it)

3. Missing in GitHub - Ask: "Create GitHub Issue for this TODO?"

   • Option A: Create issue with todo label
   • Option B: Skip

4. Missing in Code/File - Ask: "GitHub Issue has label but no local reference. What to do?"

   • Option A: Remove the todo:code or todo:file label from GitHub
   • Option B: Skip (leave as is)

Example interaction:

Found 1 inconsistency:

[1] Status Mismatch: todo/todo_open.md:5
    Issue #16: file says 'investigating' but GitHub has 'in-progress' label

    How to resolve?
    A) Update file to 'in-progress'
    B) Remove 'in-progress' label from GitHub
    C) Skip

/todo search [query] [--label <label>]

Search GitHub Issues by keyword and/or label filter.

Usage:

/todo search animation          # Search for "animation" in title/body
/todo search --label bug        # Filter by label
/todo search gui --label high   # Combined: keyword + label
/todo search --label in-progress  # Find all in-progress issues

Implementation:

# Keyword search (searches title and body)
gh issue list --label todo --search "animation" --state open --json number,title,labels,state

# Label filter (can use partial match)
gh issue list --label todo --label "priority:high" --state open --json number,title,labels,state

# Combined
gh issue list --label todo --label bug --search "animation" --state open --json number,title,labels,state

# Include closed issues
gh issue list --label todo --search "animation" --state all --json number,title,labels,state

Output format:

=== Search Results: "animation" ===

| #   | State  | Labels                  | Title                    |
|-----|--------|-------------------------|--------------------------|
| #45 | open   | bug, priority:high      | GUI Animation Bug        |
| #12 | closed | enhancement             | Animation speed control  |

Found 2 issues. Use `/todo start #id`, `/todo done #id`, or `/todo analyze #id` to operate on results.

Common label filters:

• --label bug - Bug issues
• --label enhancement - Feature requests
• --label priority:high / priority:medium / priority:low - By priority
• --label in-progress - Currently being worked on
• --label analyzed - Already analyzed by Claude
• --label todo:code - From code comments
• --label todo:file - From todo files

Options:

• --all - Include closed issues (default: open only)
• --limit N - Limit results (default: 30)

/todo track

For each untracked code TODO:

1. Assign next available ID (check existing TC# numbers)
2. Create GitHub Issue with todo and todo:code labels
3. Update code comment with issue number: # TODO [#123]: text

/todo analyze [#id]

Read the code context around a TODO, update the GitHub Issue description with analysis, and add analyzed label.

/todo start #id

Add in-progress label to the GitHub Issue.

/todo done #id

Close the GitHub Issue. Optionally remove or update the code comment.

/todo reject #id [reason]

Add wontfix label and close the issue with reason.

ID Schema

Code Comments

Format: # TODO [ID]: description

Examples:

• # TODO [TC1]: Move single step mode into operator
• # TODO [#45]: Fix animation bug
• # CLAUDE [#123]: Review this logic

GitHub Labels

Status Flow

new → analyzed → in_progress → completed
                     │
                     └──────────▶ rejected (wontfix)

Todo Files

Files with "todo" in filename are tracked:

• Should be moved to todo/ folder
• Files with "new entries" section need processing
• Each entry should become a GitHub Issue

Processing New Entries

1. Find files with has_new_entries: true in scan output
2. Read the new entries section
3. Create GitHub Issues for each entry
4. Update file to mark entries as processed

Token-Saving Strategy

1. Always run Python script first - It does the file scanning
2. Use --json flag for programmatic parsing: python .claude/skills/todo/todo_scan.py --json
3. Only read specific files when analyzing individual TODOs
4. Cache GitHub queries - The script already queries once

Migration Notes

Existing tracked items use these ID formats:

• TC1-TC6 - Code TODOs
• B#, G#, A#, Q#, S#, D# - Categorized tasks in todo_open.md

These will be migrated to GitHub Issues when /todo track is run.

Important

• NEVER create issues without user approval - Show what will be created first
• Ask before modifying code - Confirm before updating TODO comments with IDs
• Preserve existing IDs - Don't reassign TC1-TC6 to new numbers