evolve-loop
Autonomous /goal loop that rethinks a project's design and adds features and improvements that can change its behavior, each change recorded and proven. Start it only when the user names evolve-loop.
- License 许可
- Apache-2.0
- 最近更新
- 2026-11-14
- Stars
- ★ 1.6k
- Forks
- 491
01 前言 Front Matter
- name
- evolve-loop
- description
- Autonomous /goal loop that rethinks a project's design and adds features and improvements that can change its behavior, each change recorded and proven. Start it only when the user names evolve-loop.
- argument-hint
- [theme or focus] [branch=...] [limits]
- user-invocable
- true
02 资源文件 1
- 参考文档agents/openai.yaml304 B
03 SKILL.md 正文
Evolve loop
Evaluate this project in full, rethink its design and components from first principles, and add the features and improvements that make it more useful, correct, fast, and simple. Changes can alter logic and results. Each change is deliberate, recorded, and proven. Everything else stays identical. The arguments of the request can set a theme, a focus, a branch, or extra limits.
Invoking this skill approves the work below within the unattended-run limits.
Work on a new branch, evolve/loop-YYYY-MM-DD, unless the arguments name one.
Read real data from dev or staging. Test data migrations only on disposable databases.
Build what the evidence supports. Write a proposal when a change needs the user's decision: unclear business rules, product scope, breaking changes, migrations of real data, permissions, or higher costs.
Verification first
Before changing behavior, agree with the user on the definition of done and a proven regression gate, as an autonomous run requires.
Start from what the project already has to check regressions. Use brainstorming and create-test for what is missing.
Record each intended change with before and after on a real example, the reason, who is affected, and its acceptance checks.
Work
Start from the purpose: who uses the project, its main flows, and what they need. Rethink its main components and flows: which design would you choose today, for the real requirements and data? Keep the current one when the gain does not pay for the change. Find ideas in evidence: errors and slow flows in the logs, bad data, manual steps users repeat, workarounds in the code, open issues, and what good similar projects offer. Test an uncertain idea with a small experiment before you build it.
Rank ideas by benefit for the cost, with data loss and security fixes first, and use rethink to challenge the ranking.
Work in small, verifiable changes. Keep the change list, proposals, experiment results, and discoveries in the task records.
Describe each new behavior in the docs with a real example.
Report the behavior changes, verification, proposals, and remaining questions against the agreed definition of done.
Done
The loop is done when, on its last commit:
- The gate exits 0 with only the declared differences, and every acceptance check fails on the baseline and passes now.
- The independent review of the whole branch has no open CRITICAL or HIGH finding.
- A final evaluation in a fresh context finds no new idea that the evidence supports and that can ship without the user's decision. No idea is open.
- Docs, changelog, and task records are current, and
git statusis clean.
04 同仓库更多 Skills
worktable-review-with-annotations
Create, inspect, reply to, update, or resolve situated Worktable annotations. Use when feedback belongs to a Doc section, block, text range, or HTML Doc, or when the user asks to address existing review feedback and coordinate a requested artifact edit.
worktable-find-and-synthesize-context
Find, assess, and synthesize context from Worktable Docs and Records. Use when the user asks what a workspace contains, wants workspace-backed research, needs current facts reconciled across artifacts, or requests a cited synthesis without changing workspace state.
worktable-create-or-update-docs
Create, edit, rename, or maintain narrative Worktable Docs. Use for plans, decisions, briefs, notes, research syntheses, and other durable prose when the user requests a new artifact or a safe update to an existing one.
worktable-create-or-manage-records
Organize repeated structured information as durable Worktable Records. Use for research datasets, inventories, directories, pipelines, trackers, and other collections whose items need identity, fields, filtering, links, validation, or individual updates.
worktable-create-or-manage-interactive-html
Create or improve durable visual and interactive artifacts in Worktable. Use for dashboards, trackers, calculators, mini apps, slide presentations, and other requests that benefit from a polished HTML Doc.
worktable-collaborate-threads
Read, answer, and hand off messages through durable Worktable threads with users or other authenticated participants. Use when the user asks to check addressed messages, reply to a participant, continue a thread, request another agent's help, wait briefly for a response, or watch for comments and messages sent to the agent.