meeting-notetaker
Pass
Audited by Gen Agent Trust Hub on Sep 29, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted external data in the form of raw meeting transcripts, which creates a potential surface for indirect prompt injection.
- Ingestion points: Raw transcripts, voice notes, and messy notes are ingested in
SKILL.md(Step 1). - Boundary markers: The instructions do not define specific delimiters (like XML tags) to isolate the transcript from the agent's instructions.
- Capability inventory: The skill's primary capability is generating structured Markdown text and email drafts. No dangerous tool-use or system-level capabilities are defined in the skill logic.
- Sanitization: There are no explicit instructions to sanitize or filter the input transcript for embedded commands, though it includes a confidentiality rule to ignore 'off-the-record' content.
Audit Metadata