gh-issue-to-demand-signal

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

• Third-party content exposure detected (high risk: 1.00). This skill explicitly fetches open, user-generated GitHub issues from arbitrary public repos via the GitHub REST API (see SKILL.md Steps 3–6) and requires the agent to read and use verbatim issue titles/bodies to cluster, score, and generate GTM messaging, so untrusted third‑party content can directly influence its actions and enable indirect prompt injection.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill fetches live issue data at runtime from the GitHub REST API (e.g. https://api.github.com/repos/{owner}/{repo}/issues?state=open&per_page=100&page=1), and that returned JSON (verbatim issue titles/bodies) is injected into the model's analysis prompts to directly drive clustering and messaging, so external content controls agent outputs.