ah-resolve-pr-review

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

• Third-party content exposure detected (high risk: 0.90). This skill explicitly fetches and ingests user-generated GitHub content (PR diffs, review threads and comment bodies, and linked issue text) via scripts/fetch_pr_data.py and the SKILL.md workflow and then reads and acts on those comments/suggestions to plan and make code changes, exposing the agent to untrusted third-party input that could inject instructions.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill calls the GitHub CLI at runtime to fetch PR data (including review threads and suggestion blocks) from GitHub (e.g., the PR URL https://github.com///pull/ and the GitHub API at https://api.github.com), and that reviewer-provided content is injected verbatim into plans/changes so remote content can directly control agent prompts and actions.