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 meeting transcripts which are untrusted external data. It lacks explicit boundary markers or instructions to treat transcript content strictly as data, creating a surface where instructions embedded in a transcript could potentially influence agent behavior.
- Ingestion points: The agent is instructed to scan local folders for transcript files in various formats (.txt, .md, .vtt, .srt, .docx) and read their contents for analysis (SKILL.md, Instructions section 1).
- Boundary markers: The instructions do not define delimiters (like XML tags or markdown blocks) or include 'ignore instructions' warnings to isolate the transcript content from the agent's control logic.
- Capability inventory: The skill utilizes the agent's filesystem read capabilities to access and process the content of transcripts across multiple meetings.
- Sanitization: There is no requirement or mechanism mentioned to sanitize, filter, or escape the text from the transcripts before it is interpolated into the analysis prompt.
Audit Metadata