meeting-ingestion
Pass
Audited by Gen Agent Trust Hub on Oct 10, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONDATA_EXFILTRATION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted meeting transcripts, which could contain malicious instructions. The skill mitigates this through mandatory verification gates, transcript-grounding requirements, and consistency checks against established data.
- [COMMAND_EXECUTION]: The skill utilizes local gbrain CLI tools for data management and repository synchronization. These operations are scoped to the agent's internal environment.
- [DATA_EXFILTRATION]: The skill contains defensive instructions to scan for and redact credentials and personally identifiable information (PII) before storing meeting data, preventing the accidental exposure of sensitive strings captured in transcripts.
Audit Metadata