Each task does ONE thing well.
/plugin marketplace add mnthe/hardworker-marketplace/plugin install ultrawork@hardworker-marketplaceEach task does ONE thing well.
❌ BAD: "Implement auth system"
✅ GOOD:
- "Create user model with password hashing"
- "Implement JWT token generation"
- "Create login endpoint"
- "Create signup endpoint"
- "Add auth middleware"
Each task has clear success criteria.
❌ BAD: "Make it work better"
✅ GOOD: "Response time < 200ms for /api/users endpoint"
Maximize what can run in parallel.
❌ BAD: A → B → C → D → E (serial)
✅ GOOD:
A ──┬── C ──┬── E
│ │
B ──┘ D ──┘
| Size | Scope | Example |
|---|---|---|
| Small | 1-2 files, < 50 lines | Add validation to form |
| Medium | 2-5 files, 50-200 lines | New API endpoint |
| Large | 5+ files, 200+ lines | Split into smaller |
Rule: If a task feels "large", it probably needs decomposition.
1. [Setup] Create file structure / boilerplate
2. [Core] Implement main logic
3. [API] Create endpoints/interfaces
4. [Test] Write tests
5. [Integration] Wire everything together
1. [Reproduce] Create failing test
2. [Fix] Implement fix
3. [Verify] Ensure test passes
4. [Regression] Check related functionality
1. [Test] Ensure existing tests cover behavior
2. [Refactor] Make structural changes
3. [Verify] All tests still pass
4. [Cleanup] Remove dead code
| Task Type | Good Criteria |
|---|---|
| API endpoint | Returns 200 with expected schema |
| Database | Migration runs, queries work |
| UI component | Renders without errors, handles states |
| Performance | Benchmark shows X improvement |
| Security | Passes security scan, no vulnerabilities |
❌ "Improve performance"
✅ "Reduce /api/search response time from 2s to <500ms"
❌ "Implement feature X and also handle edge case Y"
✅ Task 1: "Implement feature X"
Task 2: "Handle edge case Y" (blocked by Task 1)
❌ Task A and B both modify same file (conflict risk)
✅ Task A modifies file, Task B blocked by A
You are an elite AI agent architect specializing in crafting high-performance agent configurations. Your expertise lies in translating user requirements into precisely-tuned agent specifications that maximize effectiveness and reliability.