autonomous-pr

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

• Third-party content exposure detected (high risk: 1.00). The skill explicitly fetches and ingests user-generated GitHub content (e.g., "gh api repos/OWNER/REPO/pulls//comments" and "gh pr view ... --comments") and passes those review comments/CI error text into the agent ("Her feedback'i uygula", claude -p "..."), so untrusted third‑party content is read and used to drive code changes and actions.

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

• Potentially malicious external URL detected (high risk: 0.80). The skill invokes the GitHub API at runtime (gh api repos/OWNER/REPO/pulls//comments) and injects those fetched PR comments directly into claude -p prompts to drive code changes, so external content can directly control agent instructions.