autonomy-select
Use before starting a task on this project to choose how much autonomy the agent gets (interactive, supervised or unattended), with the recommended level and the reason; the developer can pick another level
- License 许可
- MIT
- 最近更新
- 2026-10-22
- Stars
- ★ 0
- Forks
- 0
01 前言 Front Matter
- name
- autonomy-select
- tags
- [ "autonomy", "workflow" ]
- description
- Use before starting a task on this project to choose how much autonomy the agent gets (interactive, supervised or unattended), with the recommended level and the reason; the developer can pick another level
03 SKILL.md 正文
Choose the autonomy level
Grant only as much autonomy as can be verified cheaply. Pick the level from the task, say why, and let the developer confirm or change it.
1. Score the task
| Question | If yes | If no |
|---|---|---|
| Is there a cheap pass/fail check (tests, lint, a real run in a disposable env) that covers the whole task? | up | down |
| Is every change easy to revert (one commit, no data outside the repo)? | up | down |
| Is the change small (about 100 lines, one concern)? | up | down |
| Is the spec already decided (issue with checkable boxes, no open design question)? | up | down |
Zones where the cost of a wrong answer is high. If the task touches one, the level is at most supervised, and unattended is refused unless the user says so:
- security: approvals, symlinks and temp files, SSRF, anything that runs or sends data out;
- public contract: CLI flags, config keys, shape of JSON or text output, file formats;
- state outside the repo: the qmd index, the user's HOME, processes of other sessions;
- process: merge method, branch protection, review flow;
- anything irreversible or ambiguous: stop and ask.
2. Recommend
- Mostly "no", or an open design question, or a new idea → interactive (
autonomy-interactive). - Mostly "yes" but a contract or a risk zone is touched, or the plan needs an approval → supervised (
autonomy-supervised). - All "yes", no risk zone, a check that cannot be faked, and a separate checker available → unattended (
autonomy-unattended).
3. Say it and let the developer override
Reply with: the recommended level, the two or three answers that decided it, and the risk zone if any. Ask: "Use this level, or another?" The developer may go up or down. Going up into a risk zone needs the user's explicit OK. Then follow the skill of the chosen level.
Never skip the checker: the agent that makes a change does not grade it.
04 同仓库更多 Skills
pr-review
Review a pull request on GitHub
pdf-export
Export reports to PDF documents
test-driven-development
Drives development with tests using the red-green-refactor loop. Use when implementing any logic, fixing any bug, or changing any behavior. Use when you need to prove that code works, when a bug report arrives, or when you're about to modify existing functionality. Use when asked to add tests to a function, fix a bug in it or change its behavior, even when the code is pasted inline and the request does not mention tests.
spec-driven-development
Creates specs before coding. Use when starting a new project, feature, or significant change and no specification exists yet. Use when drafting a PRD or requirements document with objectives and scope, or when requirements are unclear, ambiguous, or only exist as a vague idea. Use when a single requirement spans several independently testable capabilities and needs decomposing into a capability map of modules before specifying.
source-driven-development
Grounds every implementation decision in official documentation. Use when you want to verify an approach against the official docs before implementing it, or when you want authoritative, source-cited code free from outdated patterns. Use when building with any framework or library where correctness matters.
shipping-and-launch
Prepares production launches. Use when preparing to deploy to production, or when asking what needs to be in place before shipping. Use when you need a pre-launch checklist, when setting up monitoring, when planning a staged rollout, or when you need a rollback strategy.