sergei-mikhailov-tg-channel-reader

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 posts and discussion replies from Telegram channels (untrusted, user-generated content) via MTProto using commands like "tg-reader fetch" and "--comments" as documented in SKILL.md and uses those messages for summarization/analysis and to drive agent behavior (After Fetching / Scheduled Tasks sections), so third‑party content can influence agent actions.