87 lines
1.9 KiB
Markdown
87 lines
1.9 KiB
Markdown
# MACP Task Brief Template
|
|
|
|
**Use this template for all MACP task briefs.** Workers that receive briefs not following this structure should flag it as an issue.
|
|
|
|
---
|
|
|
|
```markdown
|
|
# <Title>
|
|
|
|
**Branch:** `feat/<branch-name>`
|
|
**Repo worktree:** `~/src/<repo>-worktrees/<task-slug>`
|
|
|
|
---
|
|
|
|
## Objective
|
|
|
|
<1-2 sentences: what is being built and why>
|
|
|
|
---
|
|
|
|
## Task 1: <Component Name>
|
|
|
|
<Description of what to build>
|
|
|
|
### Requirements:
|
|
- <Specific, testable requirements>
|
|
|
|
### Key Functions/APIs:
|
|
<Code signatures or interface definitions>
|
|
|
|
### Constraints:
|
|
- <Language, dependencies, patterns to follow>
|
|
|
|
---
|
|
|
|
## Task 2: <Component Name>
|
|
<Same structure as Task 1>
|
|
|
|
---
|
|
|
|
## Tests (MANDATORY)
|
|
|
|
**Every brief MUST include a Tests section. Workers MUST write tests before or alongside implementation. Tests MUST pass before committing.**
|
|
|
|
### Test file: `tests/test_<module>.py`
|
|
|
|
### Test cases:
|
|
1. `test_<name>` — <what it verifies>
|
|
2. `test_<name>` — <what it verifies>
|
|
...
|
|
|
|
### Test runner:
|
|
```bash
|
|
python3 -m unittest discover -s tests -p 'test_*.py' -v
|
|
```
|
|
|
|
---
|
|
|
|
## Verification
|
|
|
|
1. All tests pass: `<test command>`
|
|
2. Python syntax: `python3 -c "import <module>"`
|
|
3. <Any additional verification steps>
|
|
|
|
## Ground Rules
|
|
- Python 3.10+ stdlib only (no pip dependencies)
|
|
- Commit message: `feat: <what changed>` (conventional commits)
|
|
- Push to `feat/<branch>` branch when done
|
|
```
|
|
|
|
---
|
|
|
|
## Brief Sizing Rules
|
|
|
|
| Brief Type | Max Items | Rationale |
|
|
|------------|-----------|-----------|
|
|
| **Build** (new code) | 2-3 | High cognitive load per item |
|
|
| **Fix** (surgical changes) | 5-7 | Low cognitive load, exact file/line/fix |
|
|
| **Review** | 1 | Naturally focused |
|
|
| **Test** (add tests) | 3-4 | Medium load, but well-scoped |
|
|
|
|
The key metric is **cognitive load per item**, not item count.
|
|
- Build = construction (high load)
|
|
- Fix = scalpel (low load)
|
|
- Review = naturally focused
|
|
- Test = moderate (reading existing code + writing test logic)
|