[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-78458":3},{"id":4,"name":5,"fullName":6,"owner":7,"repo":5,"description":8,"homepage":9,"htmlUrl":10,"language":11,"languages":10,"totalLinesOfCode":10,"stars":12,"forks":13,"watchers":14,"openIssues":15,"contributorsCount":16,"subscribersCount":16,"size":16,"stars1d":17,"stars7d":18,"stars30d":19,"stars90d":16,"forks30d":16,"starsTrendScore":20,"compositeScore":21,"rankGlobal":10,"rankLanguage":10,"license":22,"archived":23,"fork":23,"defaultBranch":24,"hasWiki":25,"hasPages":23,"topics":26,"createdAt":10,"pushedAt":10,"updatedAt":27,"readmeContent":28,"aiSummary":29,"trendingCount":16,"starSnapshotCount":16,"syncStatus":14,"lastSyncTime":30,"discoverSource":31},78458,"agent-skills","openclaw\u002Fagent-skills","openclaw","Useful skills for agents and claws.","",null,"TypeScript",779,41,2,1,0,7,426,744,31,92.87,"MIT License",false,"main",true,[],"2026-06-12 04:01:23","# OpenClaw Agent Skills\n\nShared skills for coding agents that work on OpenClaw projects.\n\nThis repo is the public canonical source for common workflows such as review\ncloseout and remote validation. The goal is simple: write a workflow once,\nreuse it everywhere, and avoid hand-copying long `SKILL.md` files across every\nrepo.\n\n## Included Skills\n\n- `autoreview`: structured closeout\u002Fcode-review workflow plus helper script.\n- `crabbox`: Crabbox\u002FTestbox remote validation workflow for broad or CI-parity\n  proof.\n\nRepo-specific product skills should stay in the repo they describe. For example,\nan `acpx` usage skill belongs in `openclaw\u002Facpx`; a general review helper belongs\nhere.\n\n## Quick Start\n\nClone the repo:\n\n```sh\ngit clone https:\u002F\u002Fgithub.com\u002Fopenclaw\u002Fagent-skills.git\ncd agent-skills\n```\n\nInstall all skills into the default agent skill directory:\n\n```sh\nscripts\u002Finstall-skills\n```\n\nInstall only selected skills:\n\n```sh\nscripts\u002Finstall-skills autoreview crabbox\n```\n\nInstall somewhere else:\n\n```sh\nscripts\u002Finstall-skills --target ~\u002F.codex\u002Fskills autoreview\n```\n\nUse copies instead of symlinks:\n\n```sh\nscripts\u002Finstall-skills --mode copy --target ~\u002F.agents\u002Fskills\n```\n\nSymlinks are best for local development because changes in this checkout are\nimmediately visible. Copies are better for portable or locked-down setups.\n\n## Codex And Claude\n\nFor Codex, symlink this repo into `~\u002F.codex\u002Fskills`:\n\n```sh\nmkdir -p ~\u002F.codex\u002Fskills\nln -sfn \"$(pwd)\u002Fskills\" ~\u002F.codex\u002Fskills\u002Fagent-skills\n```\n\nFor Claude Code, symlink this repo into `~\u002F.claude\u002Fskills`:\n\n```sh\nmkdir -p ~\u002F.claude\nln -sfn \"$(pwd)\u002Fskills\" ~\u002F.claude\u002Fskills\n```\n\nIf `~\u002F.claude\u002Fskills` already points at another shared skills folder, add\nsymlinks inside that folder instead:\n\n```sh\nln -sfn \"$(pwd)\u002Fskills\u002Fautoreview\" \u002Fpath\u002Fto\u002Fshared-skills\u002Fautoreview\nln -sfn \"$(pwd)\u002Fskills\u002Fcrabbox\" \u002Fpath\u002Fto\u002Fshared-skills\u002Fcrabbox\n```\n\nRecommended one-liner for repo `AGENTS.md` files:\n\n```text\nShared agent workflows: install or symlink https:\u002F\u002Fgithub.com\u002Fopenclaw\u002Fagent-skills for `autoreview`, `crabbox`, and other common skills; do not vendor shared skills here unless this repo intentionally needs a zero-setup snapshot.\n```\n\n## Zero-Setup Repos\n\nSome important repos should work for contributors who only cloned that repo and\nnever installed shared skills. Those repos may vendor a generated snapshot under\n`.agents\u002Fskills\u002F\u003Cname>`.\n\nThat snapshot is a distribution artifact, not the source of truth:\n\n- edit canonical skills here first\n- sync snapshots downstream after review\n- keep downstream copies small in number\n- add provenance and drift checks when a repo vendors a snapshot\n\n`autoreview` is a good candidate for a zero-setup snapshot in flagship repos\nbecause review closeout is part of the contribution workflow. Large operational\nskills should be vendored only when the repo genuinely needs them available\nwithout setup.\n\n## Repository Layout\n\n```text\nskills\u002F\n  autoreview\u002F\n    SKILL.md\n    scripts\u002F\n  crabbox\u002F\n    SKILL.md\nscripts\u002F\n  install-skills\n  validate-skills\n```\n\nEach skill lives in `skills\u002F\u003Cname>\u002F` and must contain `SKILL.md`. Helper scripts\nbelong inside that skill's `scripts\u002F` directory.\n\n## Validate\n\nRun this after edits:\n\n```sh\nscripts\u002Fvalidate-skills\n```\n\nThe validator checks every `skills\u002F*\u002FSKILL.md` for YAML frontmatter plus required\n`name` and `description`.\n\n## Editing Rules\n\n- Keep descriptions short and useful for routing.\n- Keep skill bodies operational rather than essay-like.\n- Do not include secrets, private hostnames, private account IDs, or private\n  URLs.\n- Prefer helper scripts for repeatable command logic.\n- Do not update vendored downstream snapshots by hand. Update this repo, then\n  sync.\n\n## License\n\nMIT.\n","openclaw\u002Fagent-skills 是一个为 OpenClaw 项目中的编码代理提供共享技能的工具库。其核心功能包括自动代码审查 (`autoreview`) 和远程验证 (`crabbox`) 等工作流程，旨在通过编写一次即可在多个地方重用这些流程，从而避免了在每个仓库中手动复制长篇 `SKILL.md` 文件的需求。该项目采用 TypeScript 编写，并且支持通过符号链接或复制的方式灵活安装所需技能至指定目录，非常适合需要标准化代码审查和测试流程的开发团队使用。此外，它还特别适用于那些希望减少新贡献者入门障碍的重要项目，通过预设快照让开发者无需额外配置即可开始工作。","2026-06-11 03:56:51","CREATED_QUERY"]