Borrowing it
Nothing to install: this file belongs to KonghaYao/peri. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.
curl -O https://raw.githubusercontent.com/KonghaYao/peri/main/.claude/skills/auto-converge/SKILL.mdgit clone --depth 1 https://github.com/KonghaYao/periWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/konghayao/peri/auto-converge)<a href="https://agentmods.dev/skills/konghayao/peri/auto-converge"><img src="https://agentmods.dev/badge/skills/konghayao/peri/auto-converge/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/konghayao/peri/auto-converge"><img src="https://agentmods.dev/badge/skills/konghayao/peri/auto-converge.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00108 | $0.02076 |
| Opus 5 | $0.00054 | $0.01038 |
| Sonnet 5 | $0.00022 | $0.00415 |
| Haiku 4.5 | $0.00011 | $0.00208 |
Grade A, and why
auto-converge scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 9d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 118 lines — stays where its author put it; the contents beside it link to each section on GitHub.
auto-converge——让规则从写完了变成能执行
两个策略驱动——
- 外部对标：先研读高质量参考（同类工具的文档、博客、规范），提取差距和缺失项，不闭门造车
- 对抗收敛：写样例 → subagent 零容忍审查 → 量化违规数 → 修复 → 重复直到违规 < 5，规则不是改一次就对的
两条策略必须同时执行。只看参考不迭代，规则停留在理论正确但 agent 执行不走；只迭代不看参考，规则在低水平收敛但没有吸收业界最佳实践。
适用于什么
不限于 SKILL.md——任何需要 agent 稳定遵守的规范都可以收敛。常见场景——
| 用户说的 | 实际要收敛的东西 | 样例是什么 |
|---|---|---|
| "这个 prompt 写好了但 agent 返回格式老不对" | prompt 模板 | 一组 agent 的实际输出 |
| "团队的 code review checklist 没人真按那个查" | review 检查清单 | 对同一段代码的审查记录 |
| "想让 agent 写的 API 文档风格统一" | 文档写作规范 | 几篇按规范写的 API 文档 |
| "这个部署检查表每次都有项漏掉" | 部署 checklist | 模拟执行记录 |
| "agent 回复用户的语气不统一" | 回复风格指南 | agent 对话记录 |
| "这个 project README 永远写不规范" | README 模板/规范 | 按模板写的 README |
| "git commit message 格式老不一致" | commit 规范 | 一组 commit message |
| "这个 skill 写出来了但 agent 不遵守" | SKILL.md 规则 | 按 skill 写的输出 |
| "想让 agent 列出来的东西都有这个格式" | 输出格式规范 | 格式化的输出样例 |
核心模式不变——只要你能说明agent 应该遵守什么且有参考源可以学习，auto-converge 就能做。收敛对象可以是 markdown 文件、prompt 字符串、JSON schema、checklist、或者任何有规则的文本文档。
触发条件
- 用户描述了一个规范/规则但 agent 屡次违反
- 用户给了一个文件路径，说"让它能执行""让它变得可验证"
- 用户说"帮我打磨""收敛""这个写出来不 work"
如果用户只说了"agent 做得不好"但没明确规范在哪——先帮他把规范落到文件里，再收敛。收敛的前提是规范有地方写。
工作流程
第 1 步：澄清三个问题
在开始对标之前，必须确认以下三点。不问清楚就动手，收敛方向可能和用户实际需求完全错位——
- 收敛什么东西？——如果有现成文件，拿路径。如果没有，帮用户把规范写下来再收敛。文件格式不限（.md、.json、.txt、prompt 字符串都可以）。
- "agent 遵守"怎么验证？——对 commit 规范来说是生成的 message 是否合规，对 prompt 来说是 agent 的输出是否按格式，对 checklist 来说是检查项是否全部覆盖。这个验证方式决定了样例长什么样。
- 参考源是什么？——没有参考不启动。参考可以是同类项目的文档、官方规范（如 Conventional Commits）、用户认为写得好的范例、或者实际运行中成功的输出记录。外部对标是收敛的方向感来源。
第 2 步：对标差距分析
并行读目标规范和参考源，产出对标差距清单——逐项列出参考有而目标规范缺的东西，按优先级排序。差距清单不追求全面，追求可操作：每条差距必须能在第 3 步转化为一条写前红灯或一条规则补充。
第 3 步：提取写前红灯 checklist
从对标差距清单中提炼 5-10 条写前自检项，作为生成样例前 30 秒扫一眼的红灯。红灯覆盖的不是"好的写法"（那是规则层面的事），而是"写完必然会犯、审查再修、下次还会犯"的机械性违规——对 commit 规范来说是前缀格式/语言/长度，对 prompt 来说是占位符/转义/必填字段。
红灯清单嵌入目标规范的执行流程中，放在"生成输出"之前。写完再查就是抓虫，写之前查是避坑。
第 4 步：对抗收敛循环
每轮——
- 按当前规范生成一组样例——样例形式取决于第 1 步确认的验证方式（commit message、prompt 输出、代码审查记录、README 草稿等等）。数量以 5-10 个为宜，太少测不全，太多审查成本高。
- 派 subagent 审查——独立上下文，prompt 中嵌入目标规范全文（不是摘要），要求零容忍、只报违规、逐条给出位置+原文+规则+建议。审查 subagent 的默认倾向是放水——prompt 里必须写明"宁可误判也不漏判"。
- 量化——统计违规数，按类别分类。
- 决策——
- 违规 < 5 且无系统性类别（某类违规 > 2）→ 收敛，进入第 5 步
- 违规 ≥ 5 或有系统性类别 → 修复样例 + 调整规范（补红灯、强化条款、新增反例），进入下一轮
- 换场景——每轮样例换个场景/主题，避免 agent 从"按规则做"退化为"背答案"。不同场景是对规则泛化能力的压力测试。
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 9d ago First seen · 118 lines · 108 tokens per session scan A 9886b17ccabc
auto-converge is a skill published in the GitHub repository KonghaYao/peri (163 stars, last pushed yesterday), licensed Apache-2.0. It adds 108 tokens to every session and 2,076 once invoked, about $0.0005 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
qa-testing
Verify your work by actually operating the app or website you changed, instead of assuming it works. Strongly recommended whenever you build, modify, or debug a web app, website, or desktop GUI app. Drive real browsers with the agent-browser CLI and native desktop apps with the cua-driver CLI. These are installed on…
plugin-creator
Create and scaffold plugin directories for Codex with a required .codex-plugin/plugin.json, optional plugin folders/files, valid manifest defaults, and personal-marketplace entries by default. Use when Codex needs to create a new personal plugin, add optional plugin structure, generate or update marketplace entries…
imagegen
Generate or edit raster images when the task benefits from AI-created bitmap visuals such as photos, illustrations, textures, sprites, mockups, or transparent-background cutouts. Use when Codex should create a brand-new image, transform an existing image, or derive visual variants from references, and the output…
babysit-pr
Babysit a GitHub pull request after creation by continuously polling review comments, CI checks/workflow runs, and mergeability state until the PR is merged/closed or user help is required. Diagnose failures, retry likely flaky failures up to 3 times, auto-fix/push branch-related issues when appropriate, and keep…
skill-creator
Create or update a Codex skill with appropriately scoped instructions and any needed supporting resources.
codex-pr-body
Update the title and body of one or more pull requests.