Granola Meeting Notes

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

• Third-party content exposure detected (high risk: 0.70). The skill’s required workflow reads Granola’s local cache “meeting notes/transcripts/summaries” (outsider-authored if those notes/transcripts were produced by someone other than the operating user) and then returns them to the agent in LLM-readable prose via commands like granola.js show ... --transcript, so indirect prompt injection from cached meeting text is possible.