git-commit

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.80). The skill's mandatory workflow (step 6 in SKILL.md) instructs the agent to actively read a public GitHub raw file (commit-types.ts at https://raw.githubusercontent.com/ruan-cat/monorepo/dev/configs-package/commitlint-config/src/commit-types.ts) using Read/WebFetch, which ingests untrusted third‑party content that directly influences commit-type/emoji selection and thus agent actions.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 1.00). The skill explicitly instructs the agent to "主动查阅" and use Read/WebFetch at runtime to fetch and rely on https://raw.githubusercontent.com/ruan-cat/monorepo/dev/configs-package/commitlint-config/src/commit-types.ts for the emoji/type list that directly governs how commit messages/prompts are constructed, making it a required runtime external dependency that controls agent instructions.