github-docs

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

• Third-party content exposure detected (high risk: 0.85). SKILL.md instructs the agent to use WebFetch at runtime to load arbitrary docs.github.com pages (public web content), whose extracted page text is then ingested into the LLM context—an outsider-authored free-text source.

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

• Potentially malicious external URL detected (high risk: 0.80). The skill explicitly instructs the agent to WebFetch docs.github.com pages at runtime (e.g., https://docs.github.com/en and other docs.github.com URLs) and to extract/use that fetched content to form answers, meaning external content fetched at runtime directly controls the agent's prompts/responses.