meeting-to-action-brief
Pass
Audited by Gen Agent Trust Hub on Sep 3, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data in the form of meeting transcripts and notes, which could contain malicious instructions intended to influence the agent's behavior.
- Ingestion points: The skill reads transcripts, summaries, notes, and stored action points from meetings (SKILL.md, Step 1).
- Boundary markers: There are no specific delimiters or markers used to isolate meeting content from the skill's instructions.
- Capability inventory: The skill does not define or use any dangerous tools such as shell execution or network access; it primarily performs text processing and refers to a 'decision-trace' tool (SKILL.md, Step 4).
- Sanitization: The skill includes instructions to paraphrase sensitive data and requires verification of the final artifact against the evidence ledger to prevent fabrication (SKILL.md, Step 2 and Step 4).
Audit Metadata