blog-analyze

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

• Third-party content exposure detected (high risk: 0.75). The skill’s runtime workflow can ingest outsider-authored free text when the user supplies a published URL: it “Fetch[es] with WebFetch, extract[s] content,” and that fetched page body becomes LLM-readable prose for Step 1 (Content Extraction) and subsequent scoring.

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

• Potentially malicious external URL detected (high risk: 0.80). The skill explicitly fetches arbitrary, user-supplied blog URLs at runtime via "WebFetch" and injects the fetched page content into the agent's analysis context, which can directly control prompts (prompt‑injection risk) — flagged: user-supplied URLs fetched via WebFetch.