retro
Skill1 starsChanged 5 days agoMITView on GitHub
复盘 agent 会话:本人要回看一次会话(默认当前这次)哪里卡住、学到什么、环境该怎么改(「复盘一下这次会话」「retro一下」「这次哪里卡了,下次怎么少绕」)。经验写进跨客户端共享记忆供所有 agent 自动召回,双向门的环境改动当场做掉,要本人定的才进待批台账。
194 other projects use a version of this.
Adopted by
Most recently taken up by
- carlos2219/web-kittheir own version, first seen Nov 15, 2026
- Gjl1u9/watch-lesstheir own version, first seen Nov 15, 2026
- Binbin-miao/workbuddy-expert-teamtheir own version, first seen Nov 15, 2026
- adelrodriguez/skillstheir own version, first seen Nov 15, 2026
- draekien/skillstheir own version, first seen Nov 15, 2026
- geoffbremner/skillstheir own version, first seen Nov 15, 2026
Used with
Projects with it also use plan-tune, ios-clean and 10 more.
Other files people use alongside this one.
Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
ShreyRaviSkillplan-tune
Self-tuning question sensitivity + developer psychographic for gstack (v1: observational). Review which AskUserQuestion prompts fire across gstack skills, set per-question preferences (never-ask / always-ask / ask-only-for-one-way), inspect the dual-track profile (what you declared vs what your behavior suggests), and enable/disable question tuning. Conversational interface — no CLI syntax required.
Also in 9 projects that use this
Saqlain-811Skillios-clean
Remove the DebugBridge SPM package and all #if DEBUG wiring from an iOS app. (gstack)
Also in 5 projects that use this
Saqlain-811Skillios-fix
Autonomous iOS bug fixer. (gstack)
Also in 5 projects that use this
seek888Skillpair-agent
Pair a remote AI agent with your browser. One command generates a setup key and prints instructions the other agent can follow to connect. Works with OpenClaw, Hermes, Codex, Cursor, or any agent that can make HTTP requests. The remote agent gets its own tab with scoped access (read+write by default, admin on request). Use when asked to "pair agent", "connect agent", "share browser", "remote browser", "let another agent use my browser", or "give browser access". (gstack) Voice triggers (speech-to-text aliases): "pair agent", "connect agent", "share my browser", "remote browser access".
Also in 11 projects that use this
ShreyRaviSkilllanding-report
Read-only queue dashboard for workspace-aware ship. Shows which VERSION slots are currently claimed by open PRs, which sibling Conductor workspaces have WIP work likely to ship soon, and what slot /ship would pick next. No mutations — just a snapshot. Use when asked to "landing report", "what's in the queue", "show me open PRs", or "which version do I claim next". (gstack)
Also in 8 projects that use this
timurgaleevSkilldocument-generate
Generate missing docs for a feature, module or project as tutorials, how-tos, reference and explanation.
Also in 7 projects that use this
- +1 more from Saqlain-811
- +2 more from ShreyRavi
Other versions194 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 194
- mattpocock/skills★ 309ksame name
- garrytan/gstack★ 141ksame name
- formbricks/formbricks★ 13ksame name
- ArcReel/ArcReel★ 6.3ksame name
- vinvcn/mattpocock-skills-zh-CN★ 5.5ksame name
- withcoral/coral★ 4.9ksame name
- bergside/awesome-design-skills★ 2.9ksame name
- MiniMax-AI/minimax-code★ 2.2ksame name
- DenchHQ/DenchClaw★ 1.7ksame name
- DenchHQ/ironclaw★ 1.7ksame name
184 more not listed
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository13 other files it gives its agents.
- architecture-diagramSkill
- arenaSkill
- diagnosing-bugsSkill
- grillSkill
- karpathy-guidelinesSkill
- memory-routingSkill
- owner-modeSkill
- skill-pruningSkill
- universal-model-promptingSkill
- upstream-prSkill
- wait-whatSkill
- windows-shellSkill
- writing-for-agentsSkill
The file
Files
本人要一次复盘:从一次会话的实际经过里收两样东西。经验是下次做同类事能直接用的做法、本人的口味、软件现状,写进跨客户端共享记忆(如 OpenViking 或共享知识目录),所有 agent 以后都能召回。环境改动(检查、文档、指针、工具)按「分诊」由 agent 定的双向门当场做掉,本人定的或单向门进待批台账。目标是本人只看真正要他拍板的那几条。
原型改写自 Matt Pocock skills 的 retro 与 chief-of-staff(MIT)。
步骤
- 读写法:调用
writing-for-agents技能。候选里拟写的文字按它写。 读原始记录。本人没指定就读当前会话。指定了旧会话就按
memory-routing技能找(如deja last --since 3d、deja <2-4 个确切词>定位,或读取对应客户端在~/.claude、~/.codex、~/.pi下的本地会话 JSONL;派过子代理且卡点在子代理内时,顺带读那个异常子代理的记录)。读完的标准:知道这次会话想做什么、在哪几处真实卡过(报错、重试、绕路、本人纠正、一次很贵的调用、找东西找了很久)。若部署了跨客户端记忆库(如 OpenViking),再对每处卡点用
ov find "<卡点里的确切词>" -u viking://~/memories -n 6查一遍,得到三样东西:- 前科:历史事件里同类事出现过几次,作为分诊「值不值得固化」的依据。
- 漏召回:如果早有一张卡讲过这个坑,这次还是踩了,说明那张卡没被召回或标题/首段写得不够准。这条要记成候选(类「记忆」)。
- 自动沉淀:记忆库已经从本次会话自动抽了哪些记录,第 6 步不要再重复写;抽错的事实顺手修正。
找候选。每处真实卡点对照下面九类,一处可以落进多类,也可以哪类都不是(一次性的事、真正的外部故障)。每条候选必须锚定到会话里的具体时刻:工具调用(工具名加关键参数或报错原文)或本人原话,带时间或消息序号。锚不住的丢掉,不凭印象提。先比较删旧规则、补一行入口指针、改用原生能力,最后才考虑加检查或钩子。
类 看什么 落点 导航 找文件、找信息花了很久;文件之间有隐藏依赖 AGENTS 的指针(一行"做 X 读 Y") 自动检查 本次实际发生的错误能被低开销确定性检查抓到。先查已有的检查命令(CI、钩子、lint 脚本),已有但没接上或坏了的才是发现;高误报或高延迟的伪一劳永逸检查不进热路径,交周期审查 检查或钩子 / 周期审查 审阅标准 审阅时漏了错。机械的做成检查;只有要判断的才写进审阅时读的文档(如 REVIEW.md)检查或钩子 / 审阅时才读的文档 常驻指令过大 常驻的 AGENTS/CLAUDE.md、全局指令里有该挪去检查或按需文档的条目 检查或钩子 / 审阅时才读的文档 工具开销 很贵的调用(大输出、反复重试、整文件读)能用更省的工具或参数代替 工具 无效指令 常驻指令里有不改变行为的句子(模型默认就这么做) AGENTS 的指针(删掉或改写) 信息获取 关键信息 agent 拿不到(日志、平台只读接口、真实状态) 工具 技能路由 该加载的技能没加载,或加载了不该加载的 技能简介修订 + 真实原话路由回归用例 记忆 有卡却没召回(第 2 步查出的漏召回);或本次把跨 agent 都该知道的事写进了某个客户端的私有记忆( ~/.claude/projects/*/memory/等),别的 agent 读不到共享记忆卡:改标题与首段,或把私有记忆迁入共享层(第 6 步) 落点与谁来定按下文「分诊」。完成标准:每处卡点都判过,要么成了候选,要么写一句为什么不提。
写候选,按严重程度排序。每条:
### <一句话标题> - 锚点:<会话号前 8 位 · 时刻或消息序号 · 工具调用或原话片段> - 类:<九类之一或几类> → 落点:<检查或钩子 / 审阅时才读的文档 / AGENTS 的指针 / 工具>:<具体文件或命令> - 门:<单向或双向>;波及面:<一个词> - 拟改:<改哪个文件、加什么或改成什么,一两句> - 采纳后:<大白话一句:agent 以后会有什么不同>门:双向门是改了能轻易撤回的(加一条检查、加一行指针);单向门是撤回代价大或影响全局行为含义的(改全局指令措辞、删文件、改调度)。波及面一个词,例如
本项目/全部 agent/一个脚本。处置。每条候选按「分诊」的「谁定」走一路:
agent 定的双向门:当场做掉。 照落点所在仓库的规则改、跑它的检查、提交;提交信息首行说改了什么,正文写锚点和
retro,这就是记录,不另入账。几条互不依赖时,各派一个全新上下文的子代理并行做,简报只给锚点、拟改和验收命令;改同一文件的串行做。检查不过或一轮修不好,就回退这条改动,转走下一路,入账时写明卡在哪。本人定的或单向门:入账,按改动的层写入对应待批台账:
- 项目层(当前项目仓库内的规则、架构、脚本):写入该项目的改进台账或
proposals.md/ Issue。 - 全局通用层(全局指令、共享技能、多客户端配置):追加到通用治理仓的
proposals.md「## 待批」,写明来源retro、锚点、理解、拟改、门与波及面、采纳后效果与回退办法。 - 未设台账的独立目录:直接在回复里列出待拍板项。
完成标准:每条候选有提交号、台账编号之一,或写明为什么都不做;编号贴在回复里。
- 项目层(当前项目仓库内的规则、架构、脚本):写入该项目的改进台账或
沉淀跨客户端共享记忆。把换一个任务还用得上、读代码和文档看不出来的经验写进所有 agent 共用的记忆层(以 OpenViking
viking://~/memories/为例;未装独立记忆库时写入~/.agents/memories/或仓库文档):内容 放哪 本人的口味和纠正(怎么沟通、怎么交付、什么样算好) 并进 preferences/下对应的规范卡,一条一行某个软件或系统的现状(装了什么、在哪、怎么还原) 并进 entities/<类>/<名>.md实体卡;被推翻的旧状态直接改掉做某类事的做法与坑(踩过的、走通的) patterns/<主题>.md经验卡只属于某一个代码仓库的规则 直接写进那个仓库的 SSOT 文档 - 并卡只留现状:先检索同主题已有卡,读全文后合并覆盖,被推翻的旧做法直接改掉,不叠历史补丁。
- 私有记忆迁入共享层:本次写进单一客户端私有目录、但其他客户端也该知道的条目,并入共享层后删掉私有副本,只留一处真相。
- 召回实测(针对向量记忆库):经验卡标题写触发场景(「做 X 时」),把下次会用来提问的词放进标题和首段。每张新写或改过的卡,用 2–3 种下次真会问的问法在全量记忆范围实测召回(不只在子目录测,防止被大量短期事件挤出 top-K);排不进召回窗口时,把长卡按问法拆成聚焦短卡。密钥与个人隐私绝不进卡。
- 回复本人:用人话简短说明共享记忆里写了或改了哪些卡(下次什么场景会用上)、当场做掉了哪几条双向门(下次少绕什么,附提交号),以及入账等拍板的几条(附编号);完整的锚点、分类、门与波及面留在提交信息和台账里,不在聊天里刷屏。
分诊
- 谁定:
owner-mode与全局指令「少请示」列的本人的域、规则含义变化与单向门,本人定;其余 agent 自己定,不请示。 - 值不值得固化:先用
git log --grep或会话检索查同类出现过几次。只出现一次、手动一下就完的,当场处理,不立规则、不留待办。 - 放哪:低开销且零误报的确定性事实做检查或钩子(写明防哪次真实错误与未来升级后的退役条件,拦下时报错直接指明替代做法,并用本次真实错例验红),能自动修就自动修;修好的事故能写成确定性回归用例的同时加一条;有误报或需要上下文判断的规则交周期审查,不放热路径;要判断且处处适用的进常驻指令,同时删掉或并掉一条旧的;只在某类任务里用的进技能或按需文档,常驻处最多留一行指针。
- 减法:新增的规则、检查、钩子、技能、脚本、流程,都必须能说出替掉了什么、为什么不能并进现有机制;说不出来就不加。钩子和检查拦不到真问题、拖慢热路径或随模型/客户端升级已过时的,直接撤掉。
边界
- 当场做的只限 agent 定的双向门。
- 只在本人主动要求复盘 agent 会话时运行(打
/retro或明说复盘本次会话)。自动定时跑的复盘容易挑无关紧要的毛病,不接定时任务、不接钩子。 - 与
diagnosing-bugs的分工:那个修眼前坏掉的代码,这个改环境让下一次不卡。
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

