instagram-replicate

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

• Third-party content exposure detected (high risk: 0.80). The skill's required workflow (SKILL.md and references/workflow.md) directly runs scripts/fetch_instagram_snapshot.py which uses yt-dlp and fetch_profile_html to download public Instagram pages (captions, comments, profile HTML, media URLs) — untrusted, user-generated third‑party content — and those fields are parsed into snapshot.json and used to control rendering/transcoding decisions (e.g., media kind, audio muxing, what to display), so the fetched content can materially influence tool actions.