agentleFS
Sign inStart free
gaoyu06gaoyu06/harness/AGENTS.md

harness

AGENTS.md3 starsChanged 3 days agoMITView on GitHub

Harness engineering 实践仓库：收集 agent 开发中可复用的经验。 规范见 harness/SPEC.md。会话开始： 维护：默认不写 note（写入条件见 SPEC）；proposing 需用户决策时停下问一个问题；结论进了 spec/ 或代码即归档 note（git mv 进 archive/ 并删 INDEX 行）；并入、归档、清 INDEX 由后台子代理按 .harness/wrap-up.md 执行，主会话不等待；会话产物放 .harness/local/；实现 note 的 commit 引用其 slug。 见 .harness/spec/standards.md。高频义务单独点名：

The file

Create a free account to mount
Files
AGENTS.md

AGENTS.md

Harness engineering 实践仓库：收集 agent 开发中可复用的经验。

结构

  • harness/ — .harness/ 状态层规范（SPEC.md）与安装模板（template/）
  • .harness/ — 本仓库自身的状态层（dogfood）
  • skills/ — 完整的 agent skills，每个子目录含 SKILL.md 与配套文件
  • prompts/ — 一次性提示词片段，不固化为 skill
  • rules/ — 全局规则文件（跨宿主适用的 rule、全局 CLAUDE.md 类）
  • hooks/ — Claude Code hook 脚本，由 install.sh 写入 settings.json
  • notes/ — 实践笔记
  • snippets/ — 代码片段与配置样例

.harness

规范见 harness/SPEC.md。会话开始：

  1. 读 .harness/notes/INDEX.md（命中才打开 note）和 .harness/spec/index.md（命中才读对应 spec）。
  2. git status——脏工作区或进行中分支即未完成任务的痕迹，向用户确认处置。
  3. 索引未命中但怀疑有历史：rg 扫 .harness/（含 archive/），仍无即视为无。

维护：默认不写 note（写入条件见 SPEC）；proposing 需用户决策时停下问一个问题；结论进了 spec/ 或代码即归档 note（git mv 进 archive/ 并删 INDEX 行）；并入、归档、清 INDEX 由后台子代理按 .harness/wrap-up.md 执行，主会话不等待；会话产物放 .harness/local/；实现 note 的 commit 引用其 slug。

约定

见 .harness/spec/standards.md。高频义务单独点名：

  • 写或改 skills/ 里的 skill：按 skills/writing-for-agents/ 的方法执行
  • 中文编写的 skill：成稿后过 skills/zh-check/
  • 写 README.md 等门面文档：只写当前成立的稳定状态；改名、分叉、迁移这类变更说明归 commit message / note，不进正文

References

  • mattpocock/skills — Matt Pocock 的 skill 集，本地 clone 于 ~/dev/skills-mp。参考点：user-invoked / model-invoked 二元划分、grill 追问原语、共享语言词汇表、spec → tickets → implement 流水线
  • aicare 工作区（~/dev/aicare）— agent-notes 与 spec 语料模式来源

Download now (ZIP)

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
What did you use it for? e.g. Reviewing pull requests in a Rails app

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.

Check your setup, free