Evaluate an Agent Skill bundle for structural integrity, trigger quality, artifact improvement, script correctness, safety, installed-tree integrity, and target
复制下面这句话,粘贴给 Claude Code、Codex、Cursor 等 AI 编程工具,它会读取安装说明并在你确认后完成安装。
请阅读 https://ai.atlankj.com/install/asset/gh-skill-release-gate-418bdf85abd8 ,按照其中的说明把「skill-release-gate」安装到你(当前 AI 工具)中。执行前先告诉我将运行的命令和写入的位置,等我确认。
查看 AI 将读取的安装说明正在读取 GitHub 原文…
内容来自 GitHub 原始文件,由原作者维护。在 GitHub 查看
Use this skill before publishing or distributing an Agent Skill directory bundle.
SKILL_ROOT to the absolute directory containing this installed
SKILL.md. Do not assume the process cwd is the installed bundle.TARGET_ROOT from the original workspace working directory and
resolve the user-supplied candidate as an absolute TARGET_BUNDLE.references/eval-contract.md from SKILL_ROOT.evals/cases.json under TARGET_BUNDLE.evals/artifacts.json under TARGET_BUNDLE.evals/evidence.json under TARGET_BUNDLE.assets/hosts.json under TARGET_BUNDLE and verify the target file hashes
against its assets/manifest.json.localEvidenceReady, but locally recomputable hashes do not prove capture.evidenceRoot matches the report,
plus the SHA-256 of its exact bytes from a separate trusted policy or
release channel. The attestation must be a regular file outside the target
bundle.scripts/evaluate_skill.py under SKILL_ROOT. For the shipped lesson
fixture, build argv from python3, that absolute evaluator path,
--fixture-demo, and the absolute TARGET_BUNDLE. For production, use the
same installed script with --attestation,
--trusted-attestation-sha256, and the absolute TARGET_BUNDLE, without
--fixture-demo.checksPassed, fixturePassed, localEvidenceReady,
trustAnchorValid, productionReady, and passed with the evidence root,
evaluation modes, failed checks, precision, recall, every raw trigger
observation, per-case repeated-run rates, artifact comparison, script and
safety evidence, installed-tree verification, and portability matrix.
Include the resolved script path, resolved target path, cwd, exact argv,
and exit code. Mark unavailable observations unverified.Return the complete JSON evaluation report. Preserve every layer-specific check and its evidence so a passing aggregate cannot hide a routing, artifact, script, safety, installed-tree, or portability failure. fixturePassed reports a successful teaching fixture. localEvidenceReady reports only local digest integrity. passed is true only when productionReady also has a valid out-of-bundle trust anchor.
If configuration is invalid, provenance is absent or mismatched, the trusted attestation is missing or invalid, a file hash differs, a required capability is absent, or any production gate fails, stop with a nonzero result and report the failed layer. The explicit --fixture-demo path may exit successfully only when fixturePassed is true, and it never makes a release claim. Never publish, install elsewhere, repair evidence, create the trust decision, or weaken a threshold automatically.
Do not publish a bundle merely because SKILL.md parses or one positive prompt activates. Do not label a package portable when a target drops required companion files or ignores required runtime extensions.