Harness agents md
Open-source Agent Skills for safe, evidence-based software delivery
npx -y skills add RayJason/rayjason-skills --skill harness-agents-mdAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
2 things to look at
- 12 days oldThe repository was created 12 days ago. New is not bad, but a brand new repository carrying a familiar-sounding name is the shape a typosquat arrives in, and there has been no time for anyone else to find a problem with it.
- 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
What its author says it does
Copied from the file, not written here
创建、审计和优化全局与项目级工程化的 AGENTS.md，覆盖作用域、协作、Git、验证与交付。
SKILL.md
3.3 KB, ~1.1k tokens by cl100k_base, as published. Nobody here has run it
驾驭 AGENTS.md
分析全局与项目级现有的 AGENTS.md 文件，整理零散、冲突、不可执行、过时的约定。先读取真实环境和仓库，再给方案；不套用通用模板。
安装
npx skills add RayJason/rayjason-skills --skill harness-agents-md --global
- 可用
bunx skills替代npx skills - 去掉
--global可安装到当前项目。
使用方法
在目标仓库中对 Agent 说：
请使用 harness-agents-md，分析全局与项目级 AGENTS.md 文件，给出建议，等我选择治理范围后再修改并验证。
工作流
- 识别宿主能力、全局指令入口、项目根目录、嵌套规则、权限和作用域；自行判断各层文件是否存在，不要求用户说明。
- 分别读取全局层与从项目根到当前目录的规则链；项目级还要检查默认分支保护、README/CONTRIBUTING 约定和既有 tasks/roadmap/进度文档。
- 先报告两层现状、冲突、重复、缺失和错放规则，再分别给出建议；相似规则以用户已有约束为准，兼容时合并去重。
- 让用户选择只改全局、只改项目、两者都改或暂不修改，并询问是否由文档持续管理项目进度。
- 经授权后更新并验证；需要分支工作流时，每个分支只处理一件事，不直接提交到受保护的默认分支。
按需加载
- 全局 + 项目级：宿主加载链、指令分层与适配：
references/agent-compatibility.md和references/agents-guidance.md. - 多 Agent 编排：是否并行、workstream/owner 拆分、主/子 Agent
职责、测试归属、集成/review 上限、发布冻结与停止条件：
references/multi-agent-workflow.md. - 项目级规则：架构边界或跨模块方案：
references/architecture-and-scope.md. - 项目级规则：任务、roadmap、进度和长期文档：
references/documentation-lifecycle.md. - 项目级规则：默认分支、分支提交、worktree 使用条件和依赖交接：
references/worktrees-and-dependencies.md. - 项目级规则：运行、发布、部署、迁移或采用证据：
references/verification-and-handoffs.md.
只加载本次需要的资料。缺少多 Agent、worktree 或自动化能力时顺序执行， 不得模拟。破坏性、生产、发布、凭证和外部数据操作必须有明确授权。
清理 worktree 时使用 scripts/cleanup_worktree.sh；默认只预览，传 --apply
前检查仓库、分支、干净状态和祖先关系。完成时报告结果、变更范围、验证证据、
跳过项、风险和 commit。
周期治理
完成一次治理后，必须询问用户是否设置 Schedule，选项包括：治理全局指令与
已安装 Skills、治理项目级 AGENTS.md/宿主适配文件、两者都治理或暂不设置。
仅在用户明确同意后创建；先确认周期、项目范围、运行位置和权限，并手动测试
任务提示词。默认建议每周、只审计并报告建议；自动改文件或清理 Skill 需要
单独授权。Git 项目需要写入时优先顺序执行；只有用户要求或隔离确有必要时
才使用 worktree。
Gives 0 of the 12 instructions most memory context skills give in ~1.1k tokens
Counted across 674 of the 847 authors here whose files we hold, read 2026-08-07
- inform the user when setup is completein 21 of 674, across 6 files
- confirm the draft with the user before writingin 21 of 674, across 7 files
- update the agent skills block in place if it existsin 21 of 674, across 6 files
- present findings to the userin 20 of 674, across 5 files
- ask the user about each decision one at a timein 20 of 674, across 5 files
- write the three docs files from seed templatesin 20 of 674, across 5 files
- edit CLAUDE.md if it existsin 18 of 674, across 3 files
- explore current repo statein 18 of 674, across 3 files
- do not overwrite user edits to surrounding sectionsin 18 of 674, across 3 files
- keep the memory index under 200 linesin 18 of 674
- back up the original file before overwritingin 16 of 674, across 8 files
- provide actionable steps and verificationin 13 of 674, across 2 files
Said here and by no other author read
- read the actual environment and repository first
- do not apply generic templates
- determine file existence independently
- read global and project rule chains separately
- check default branch protection
- report conflicts duplicates and missing rules
Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once. Length counted with cl100k_base; the agent that loads this file may tokenize it differently.