ad-angle-miner

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

• Third-party content exposure detected (high risk: 1.00). The skill explicitly instructs the agent to fetch and ingest untrusted, user‑generated content from public sites—e.g., Phase 1's Review Mining using Apify/Amazon and web_search for G2/Capterra/TrustRadius, Reddit scraping via Apify, and Twitter/X and competitor ad library web_search—and then requires the agent to read and act on that content to extract and rank ad angles, so third‑party content can materially influence tool use and decisions.

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

• Potentially malicious external URL detected (high risk: 0.80). This skill invokes Apify actor endpoints at runtime (e.g., POST https://api.apify.com/v2/acts/web_wanderer~amazon-reviews-extractor/runs and POST https://api.apify.com/v2/acts/trudax~reddit-scraper-lite/runs), requires an APIFY_API_TOKEN, and thus relies on executing remote Apify actors (remote code) as a required dependency.