Loading...
Loading...
Manage project todos in todo.md files with task states (pending, in_progress, completed). PROACTIVELY invoke when Claude detects need to create or update a todo during a session. Use when breaking down work into tasks, tracking progress across sessions, organizing complex multi-step projects, or coordinating task states. Supports refactor/migration tracking with ADR integration.
npx skill4agent add dawiddutoit/custom-claude util-manage-todo./todo.md./todo.md| Use This Skill | Session Todo (rare) |
|---|---|
| Bug fixes, features, refactors | Temporary spikes only |
| Multi-step work, team coordination | Throwaway research |
| Work committed to git | Quick experiments (<1hr) |
| If unsure → use this skill |
.claude/artifacts/YYYY-MM-DD/todos/todo-{name}.md./todo.md./todo.md./todo.md# Todo: [Feature Name]
Date: YYYY-MM-DD
## Objective
[Clear statement of goal]
## Tasks
### Task 1: [Title]
**Status:** 🔴 Not Started | 🟡 In Progress | 🟢 Complete
**Priority:** High | Medium | Low
**Description:**
What needs to be done.
**Acceptance Criteria:**
- [ ] Functionality implemented
- [ ] Tests pass
- [ ] Code follows conventions
## Progress Summary
- Total: X | Completed: Y (Z%)| State | Meaning |
|---|---|
| 🔴 Pending | Not started |
| 🟡 In Progress | Actively working |
| 🟢 Complete | Acceptance criteria met |
| ⚫ Blocked | Waiting on dependency |
| Phase | What It Proves |
|---|---|
| CREATION | Artifact exists |
| CONNECTION | Wired into system |
| VERIFICATION | Works at runtime |
### Phase 1: CREATION
- [ ] Create [module]
- [ ] Unit tests pass
### Phase 2: CONNECTION
- [ ] Import in [consumer]
- [ ] Register in [system]
- [ ] Verify: `grep "module" src/`
### Phase 3: VERIFICATION
- [ ] Integration test passes
- [ ] Execution logs attached
- [ ] Expected outcome observedtodo.mdUpdate todo.md → Update TodoWrite → Verify syncvalidate-refactor-adr| Anti-Pattern | Why Bad |
|---|---|
| Todos for trivial tasks | Overhead exceeds value |
| Vague descriptions | No clear acceptance criteria |
| Refactor without ADR | Violates policy |
| TodoWrite out of sync | Breaks single source of truth |
| Tasks too large/small | Poor granularity |
| Session todo for production work | Should use ./todo.md |
| Metric | Target |
|---|---|
| Context reduction | 80% vs agent prompts |
| Sync accuracy | 100% todo.md ↔ TodoWrite |
| ADR compliance | 100% for refactors |
| Integration verification | All Phase 3 tasks have evidence |