ci-cd-supply-chain-audit

English

Purpose

Audit CI/CD and software supply-chain risk.

Workflow

1. Inspect workflows and build scripts.
2. Check token permissions.
3. Check secrets exposure.
4. Check third-party actions and dependency changes.
5. Check release artifacts and publishing.
6. Output findings and recommended gates.

Safety rules

Do not run publish, deploy, release, or credentialed commands.

Canonical finding format

id: F-001
severity: Critical | High | Medium | Low | Informational
confidence: High | Medium | Low
category:
affected_code:
root_cause:
exploit_path:
preconditions:
impact:
evidence:
minimal_fix:
regression_test:
auto_fix_suitability: Safe | Needs Human Review | Do Not Auto-Fix
notes:

v0.6 operational guardrails

• Keep the skill within its stated trigger conditions and the user's explicitly provided scope.
• Preserve project safety boundaries: audit-only by default; Do not execute exploits, Do not auto-merge, Do not upload private source code or secrets, and do not scan unrelated repositories without explicit user request.
• Ask for explicit human approval before patching high-risk auth, IAM, governance, funds, terminal, or agent-tooling behavior.
• Report validation performed, files changed, residual risk, and any skipped future-phase work when finished.

中文

目的

使用这个 skill 进行CI/CD 与供应链审计。它应该帮助审查者把输入边界、风险证据、影响、修复建议和回归测试组织成可复核的安全输出。

触发条件

适用于 workflow、dependency、build script、release、artifact、package publishing 和凭据暴露风险。如果请求超出这些边界，先说明范围差异，并选择更合适的 prompt、skill 或人工 review 路径。

不适用场景

不要用于普通 runtime authz、智能合约协议审计或法律合同 review。不要把这个 skill 当作自动扫描整个仓库、执行 exploit、上传私有源码或 secrets、自动提交、自动推送或 auto-merge 的许可。

操作流程

1. 明确用户给出的目标、允许查看的材料和不能触碰的范围。
2. 收集必要上下文，但只读取完成任务所需的文件、diff、workflow、fixture 或文档。
3. 识别 trust boundary、privileged operation、sensitive data、preconditions 和 security impact。
4. 只报告有 evidence 的 finding；缺少上下文时写 question 或 assumption。
5. 为 confirmed issue 提出 minimal fix，并规划workflow permission、lockfile、secret scanning、artifact path 和 release dry-run gate。
6. 完成后报告验证输出、残余风险和需要人工确认的事项。

安全规则

默认 audit-only。未经明确授权，不 patch、不 commit、不 push、不创建 PR、不 merge。不要执行 exploit，不要访问生产系统，不要打印 secrets。涉及 IAM、authz 模型、资金、治理、terminal 执行或 agent-tooling 权限的修复必须进入人工 review。

输出要求

使用 canonical finding format。每个 finding 都要包含 severity、confidence、category、affected_code、root_cause、exploit_path、preconditions、impact、evidence、minimal_fix、regression_test、auto_fix_suitability 和 notes。