meeting-minutes-taker
Pass
Audited by Gen Agent Trust Hub on Sep 24, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process meeting transcripts, which are untrusted external inputs. This creates a surface where a transcript could contain malicious instructions designed to manipulate the agent's summary or the retrieval self-test logic.\n
- Ingestion points: Meeting transcripts are read from the local file system during the analysis phase (Step 1).\n
- Boundary markers: The instructions lack specific boundary delimiters or instructions for the model to ignore potential directives embedded within the transcript text.\n
- Capability inventory: The skill uses the
Tasktool to spawn subagents and writes multiple files to a local directory.\n - Sanitization: There is no explicit sanitization or filtering of transcript content before it is processed.\n- [DYNAMIC_EXECUTION]: The skill relies on the
Tasktool to launch multiple subagents with prompts that are generated at runtime. This creates a dynamic execution chain where subagents process untrusted transcript data based on prompts constructed by the parent agent.
Audit Metadata