contributor

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

• Third-party content exposure detected (high risk: 0.85). Outsider-authored free text is ingested via gh issue view <number> -R <owner>/<repo> --json body,comments (issue body and comment thread written by other contributors/maintainers), which the workflow then “deep-reads” and extracts into the agent’s LLM context.

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

• Potentially malicious external URL detected (high risk: 0.70). The skill instructs cloning a user-specified GitHub repo at runtime (gh repo fork / --clone --remote) and then running the project's setup and test commands, which causes execution of remote code fetched from that repository.