multi-agent-meeting
Pass
Audited by Gen Agent Trust Hub on Oct 26, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The templates 'assets/meeting-templates/customer-insight-analysis.md' and 'assets/meeting-templates/product-feature-review.md' instruct the agent to execute local shell commands, specifically 'python scripts/customer_interview_analyzer.py' and 'python scripts/rice_prioritizer.py'. These scripts are not part of the provided skill files, making their logic and safety unverifiable during analysis.\n- [INDIRECT_PROMPT_INJECTION]: The skill provides a workflow for analyzing external interview transcripts, which introduces a vulnerability to instructions embedded in that data.\n
- Ingestion points: The agent is instructed to process content from 'interview_transcript.txt' or user-supplied text blocks in the 'Customer Insight Analysis' template.\n
- Boundary markers: There are no explicit delimiters or instructions telling the agent to treat the transcript content strictly as data rather than instructions.\n
- Capability inventory: The skill allows for command execution (via referenced Python scripts), file generation (creating PRD drafts), and impacting subsequent multi-agent discussions.\n
- Sanitization: The skill does not provide mechanisms for sanitizing or validating external input before processing it.
Audit Metadata