meeting-insights-analyzer
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external meeting transcripts which are untrusted data sources. This creates a surface where an attacker could embed malicious instructions within a transcript to manipulate the agent's behavior. \n- Ingestion points: Reads meeting transcript files (e.g., .txt, .md, .vtt, .srt, .docx) from local folders as described in the 'Instructions' section. \n- Boundary markers: The instructions lack explicit delimiters or negative constraints directing the agent to ignore any instructions embedded within the processed text. \n- Capability inventory: The agent utilizes file system read capabilities across user-specified directories to retrieve and process transcript content. \n- Sanitization: There are no procedures defined for sanitizing or escaping the content of the transcripts before they are interpolated into the analysis prompt.
Audit Metadata