fathom

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

• Third-party content exposure detected (high risk: 0.90). The SKILL.md explicitly instructs the agent to fetch meeting transcripts and summaries from the third‑party Fathom API via endpoints like https://gateway.maton.ai/fathom/external/v1/recordings/{recording_id}/transcript and /summary, which return user-generated meeting content that the agent is expected to read and could therefore contain untrusted instructions influencing behavior.

HIGH W008: Secret detected in skill content (API keys, tokens, passwords).

• Secret detected (high risk: 1.00). I flagged the webhook secret because it is a high-entropy, literal secret present in the example response: "whsec_x6EV6NIAAz3ldclszNJTwrow". It is not a placeholder (e.g., YOUR_API_KEY) or redacted value and looks like a real webhook signing secret.

Ignored items:

• MATON_API_KEY / "YOUR_API_KEY" — placeholder per docs.
• session_token shown as "..." — redacted.
• UUIDs (e.g., connection_id: 21fd90f9-...) and base64 cursors (eyJwYWdlX251bSI6Mn0=) — not secrets.
• Example/simple passwords are not present beyond placeholders.

Therefore there is one high-entropy secret in the prompt.