Skip to content
Back to search

agent-skills

Useful skills for agents and claws.

Safety scan

A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.

Checked today

Check it deeper on your machine

It has changed since its scan on 11 Nov 2026.

As a Claude Code plugin, install it and then ask Claude Code:

/plugin marketplace add keithmackay/mackayi/plugin install scanskillsafety@mackayiDo a deep scan of https://github.com/openclaw/agent-skills with scanskillsafety

Or from a terminal, clone it and run it:

git clone https://github.com/keithmackay/scanskillsafety-skill ~/.claude/skills/scanskillsafetynode ~/.claude/skills/scanskillsafety/cli.cjs --deep 'https://github.com/openclaw/agent-skills'

It checks the exact commit: every file, its packages against OSV (api.osv.dev), how it compares with this page's scan (findsafeskills.com receives the repository's address), and, if you agree, Claude reading every file with your own Claude subscription. It never runs the code.

Badge for your README
findsafeskills safety rating

Markdown

[![findsafeskills safety rating](/api/listings/openclaw--agent-skills/badge.svg)](/listings/openclaw--agent-skills)

HTML

<a href="/listings/openclaw--agent-skills"><img src="/api/listings/openclaw--agent-skills/badge.svg" alt="findsafeskills safety rating" /></a>

Install

i

Install instructions from the README

As written by the repo's author. findsafeskills has not verified these steps.

Clone the repo:

git clone https://github.com/openclaw/agent-skills.git
cd agent-skills

List available skills:

scripts/install-skills --list

Preview an install without changing files:

scripts/install-skills --dry-run

Install all skills into the default agent skill directory:

scripts/install-skills

Install only selected skills:

scripts/install-skills autoreview crabbox

Install somewhere else:

scripts/install-skills --target ~/.codex/skills autoreview

Use copies instead of symlinks:

scripts/install-skills --mode copy --target ~/.agents/skills

Replace an existing installed skill:

scripts/install-skills --force autoreview

Install targets must not contain or sit inside the source skills. The installer checks every selected destination before changing files, including with `--force` or `--dry-run`; installing directly over the same source directory is skipped.

Symlinks are best for local development because changes in this checkout are immediately visible. Copies are better for portable or locked-down setups. Node-based skills declare their module format locally, so copied helpers also run inside CommonJS projects without installing dependencies.

What was scanned

The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.

README.md · README · 292 lines

View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)

This file has changed since it was scanned, so this is not exactly the text that was scanned, and no line is marked. The findings are listed with the safety scan above.

  1. # OpenClaw Agent Skills
  2. ![Agent Skills banner](docs/assets/readme-banner.jpg)
  3. Shared skills for coding agents that work on OpenClaw projects.
  4. This repo is the public canonical source for common workflows such as review
  5. closeout and remote validation. The goal is simple: write a workflow once,
  6. reuse it everywhere, and avoid hand-copying long `SKILL.md` files across every
  7. repo. See [VISION.md](VISION.md) for catalog boundaries and admission principles.
  8. ## Included Skills
  9. - `agent-transcript`: local-only, redacted PR/issue transcript provenance.
  10. - `autoreview`: structured closeout/code-review workflow plus helper script.
  11. - `behavior-validator`: source-blind validation of user-visible behavior against
  12. a contract.
  13. - [`beam`](skills/beam/README.md): self-contained, authenticated, redacted
  14. publication of local coding sessions to a read-only OpenClaw catalog.
  15. - `crabbox`: Crabbox/Testbox remote validation workflow for broad or CI-parity
  16. proof.
  17. - `handoff`: path-free prompt handoff workflow for delegating a task to another
  18. agent.
  19. - `readme-standard`: house README structure, badge row, tone, and verification
  20. gates for steipete/openclaw repos.
  21. - `session-viewer`: local searchable HTML viewer for agent session JSONL.
  22. - `team-handoff`: start a seeded worktree session on a shared OpenClaw Gateway as
  23. yourself in one request.
  24. Repo-specific product skills should stay in the repo they describe. For example,
  25. an `acpx` usage skill belongs in `openclaw/acpx`; a general review helper belongs
  26. here.
  27. ## Quick Start
  28. Clone the repo:
  29. ```sh
  30. git clone https://github.com/openclaw/agent-skills.git
  31. cd agent-skills
  32. ```
  33. List available skills:
  34. ```sh
  35. scripts/install-skills --list
  36. ```
  37. Preview an install without changing files:
  38. ```sh
  39. scripts/install-skills --dry-run
  40. ```
  41. Install all skills into the default agent skill directory:
  42. ```sh
  43. scripts/install-skills
  44. ```
  45. Install only selected skills:
  46. ```sh
  47. scripts/install-skills autoreview crabbox
  48. ```
  49. Install somewhere else:
  50. ```sh
  51. scripts/install-skills --target ~/.codex/skills autoreview
  52. ```
  53. Use copies instead of symlinks:
  54. ```sh
  55. scripts/install-skills --mode copy --target ~/.agents/skills
  56. ```
  57. Replace an existing installed skill:
  58. ```sh
  59. scripts/install-skills --force autoreview
  60. ```
  61. Install targets must not contain or sit inside the source skills. The installer
  62. checks every selected destination before changing files, including with `--force`
  63. or `--dry-run`; installing directly over the same source directory is skipped.
  64. Symlinks are best for local development because changes in this checkout are
  65. immediately visible. Copies are better for portable or locked-down setups.
  66. Node-based skills declare their module format locally, so copied helpers also
  67. run inside CommonJS projects without installing dependencies.
  68. ## Codex And Claude
  69. For Codex, symlink this repo into `~/.codex/skills`:
  70. ```sh
  71. mkdir -p ~/.codex/skills
  72. ln -sfn "$(pwd)/skills" ~/.codex/skills/agent-skills
  73. ```
  74. For Claude Code, symlink this repo into `~/.claude/skills`:
  75. ```sh
  76. mkdir -p ~/.claude
  77. ln -sfn "$(pwd)/skills" ~/.claude/skills
  78. ```
  79. If `~/.claude/skills` already points at another shared skills folder, add
  80. symlinks inside that folder instead:
  81. ```sh
  82. ln -sfn "$(pwd)/skills/autoreview" /path/to/shared-skills/autoreview
  83. ln -sfn "$(pwd)/skills/crabbox" /path/to/shared-skills/crabbox
  84. ```
  85. Recommended one-liner for repo `AGENTS.md` files:
  86. ```text
  87. Shared agent workflows: install or symlink https://github.com/openclaw/agent-skills for `autoreview`, `crabbox`, and other common skills. Autoreview uses one shared installation; repository entrypoints refer to it.
  88. ```
  89. ## Shared Autoreview
  90. Install autoreview once with `python3 scripts/install-skills autoreview`. The
  91. default `~/.agents/skills/autoreview` symlink serves every repository. On Windows,
  92. use `python`; add `--mode copy` if symlinks are unavailable.
  93. Repositories keep only a `.agents/skills/autoreview/SKILL.md` entrypoint copied
  94. from [the repository template](skills/autoreview/references/repository-entrypoint.md).
  95. Agents read the full installed skill and run its helper from the repository being
  96. reviewed. Repo-specific review thresholds and contribution rules stay in that
  97. repository. Helpers, fixtures, and their tests stay here.
  98. Contribute shared improvements here first. After active reviews finish, update
  99. this checkout once to update every symlinked consumer. Copy-mode installations
  100. need `python3 scripts/install-skills --mode copy --force autoreview` after the
  101. source update. Review runs do not download or update code automatically.
  102. For pinned source archives, upgrade/rollback steps, and copy-versus-symlink
  103. identity limits, see [source installation](docs/RELEASING.md#installing-a-pinned-source).
  104. To compare installed helper content, run
  105. `python3 ~/.agents/skills/autoreview/scripts/autoreview --version` (use your
  106. custom install path when applicable). The output is `autoreview source-sha256:`
  107. followed by the helper's SHA-256. It does not identify a release or the full skill
  108. and does not read configuration, invoke Git or a reviewer, or write reports.
  109. ## Zero-Setup Repos
  110. Some important repos should work for contributors who only cloned that repo and
  111. never installed shared skills. Those repos may vendor a generated snapshot under
  112. `.agents/skills/<name>`.
  113. That snapshot is a distribution artifact, not the source of truth:
  114. - edit canonical skills here first
  115. - sync snapshots downstream after review
  116. - keep downstream copies small in number
  117. - add provenance and drift checks when a repo vendors a snapshot
  118. This option does not apply to `autoreview`, which uses the shared installation
  119. above. Vendor other operational skills only when the repo needs them available
  120. without setup.
  121. ## Repository Layout
  122. ```text
  123. skills/
  124. agent-transcript/
  125. SKILL.md
  126. scripts/
  127. autoreview/
  128. SKILL.md
  129. scripts/
  130. behavior-validator/
  131. SKILL.md
  132. references/
  133. beam/
  134. README.md
  135. SKILL.md
  136. references/
  137. scripts/
  138. crabbox/
  139. SKILL.md
  140. handoff/
  141. SKILL.md
  142. session-viewer/
  143. SKILL.md
  144. scripts/
  145. scripts/
  146. install-skills
  147. validate-skills
  148. ```
  149. Each skill lives in `skills/<name>/` and must contain `SKILL.md`. Helper scripts
  150. belong inside that skill's `scripts/` directory.
  151. ## Validate
  152. Development checks use Python 3.14 and Node.js 26, matching CI. Install the
  153. development dependencies once, then run the shared check command:
  154. ```sh
  155. python3 -m venv .venv
  156. . .venv/bin/activate
  157. python -m pip install -r requirements-dev.txt
  158. npm ci --ignore-scripts
  159. python scripts/check-skills
  160. ```
  161. The check command runs frontmatter validation, syntax checks, all Python and
  162. Node tests, and `npm run typecheck` for the session viewer. The Node packages
  163. are development-only; installed skills do not need `npm install`. The separate
  164. macOS CI job uses the `xcode-27` Apple Silicon runner and installs a pinned Codex CLI to exercise native sandbox access
  165. controls without a reviewer account or provider request.
  166. Python autoreview suites run through `check-skills` with a fresh temporary
  167. directory beneath the caller's home and without inherited reviewer settings or
  168. provider credentials. The caller's home must be writable and have private storage
  169. ancestry; do not point it at shared `/tmp`. Each fixture supplies its own
  170. configuration. The runner preserves executable lookup,
  171. the OS temporary directory, locale, and selected Python dependency settings,
  172. including `PYTHONPATH`; use a trusted toolchain and dependency path. This
  173. isolation does not apply to direct `python -m unittest` invocations, including
  174. the separate macOS sandbox job, or to the runner's other validation commands.
  175. On Windows, `scripts/check-skills`, its directly invoked self-test
  176. `scripts/check-skills.test.py`, and `scripts/installed-autoreview.test.py` prepare a
  177. restricted primary process when the current invocation does not meet
  178. the ordinary test profile. Invoke these files directly with
  179. Python; module (`-m`), command (`-c`), or mutated argument forms are unsupported.
  180. They retain the selected Python interpreter, flags, arguments, working
  181. directory, environment, standard streams,
  182. and exit status. With a shared console, the child handles Ctrl+C while the parent
  183. waits for its result. Failure to create or verify that process fails the test command;
  184. it does not omit tests or change the installed skill's runtime behavior.
  185. The child must have medium or lower integrity; lowering its integrity leaves the
  186. original parent token unchanged. New objects default to the child’s user as owner.
  187. The fresh token grants access to that user and SYSTEM, with matching defaults for
  188. new objects that do not receive inherited or explicit permissions. Existing source
  189. permissions are unchanged.
  190. Local/domain and Entra account identities are
  191. supported; system and virtual-service identities are not. This does not classify
  192. how domain accounts are used. The account domain's Domain Users grant is retained;
  193. other custom group grants may be removed by the restricted test profile.
  194. The full native proof suite requires permission to duplicate and adjust the
  195. current process token and create its restricted child. Missing those test
  196. capabilities fails the requested proof even when read-only admission succeeds.
  197. For the main runner and installed-source tests, when the original primary token
  198. does not meet that profile, the parent runs native bootstrap proofs before
  199. launching the child: a restricted thread cannot hide its original primary authority, a marker
  200. cannot bypass admission, a failed parent check stops execution, and a real child
  201. preserves invocation details. These auxiliary checks run noninteractively; an
  202. interactive flag still applies to the requested child, without a second prompt
  203. in the waiting parent. The restricted child runs the requested suites and
  204. the regular invocation check. These checks establish the test process boundary;
  205. they do not by themselves establish a skill's filesystem custody policy. The
  206. bootstrap retains user identity grants and does not provision accounts or
  207. rewrite file or desktop permissions. Hosts that require removed administrative
  208. grants for their interpreter or checkout must fix that access before testing.
  209. Compilation uses a disposable bytecode cache instead of writing into the source
  210. checkout. Installed-source tests trust only their exact checkout for Git metadata
  211. reads, without changing persistent Git configuration. Auxiliary proof output uses explicit
  212. UTF-8 internally and is relayed through the caller’s text streams.
  213. For a quick frontmatter-only check, run `scripts/validate-skills`. It checks
  214. every `skills/*/SKILL.md` for YAML frontmatter plus required `name` and
  215. `description` strings. Frontmatter must be a mapping enclosed by standalone
  216. `---` lines; trailing spaces or tabs are allowed on the closing line.
  217. Session exports can contain sensitive conversation data. Treat `session-viewer`
  218. HTML as local/private output unless it has been separately redacted and reviewed.
  219. ## Editing Rules
  220. - Keep descriptions short and useful for routing.
  221. - Keep skill bodies operational rather than essay-like.
  222. - Do not include secrets, private hostnames, private account IDs, or private
  223. URLs.
  224. - Prefer helper scripts for repeatable command logic.
  225. - Do not update vendored downstream snapshots by hand. Update this repo, then
  226. sync.
  227. See [docs/RELEASING.md](docs/RELEASING.md) for the source-release process.
  228. ## License
  229. MIT.

Ratings and comments

Sign in to rate this skill.

Sign in

Free to use. If it’s helped, pass it on.