multi-agent
Pass
Audited by Gen Agent Trust Hub on Sep 3, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill defines a 'Drone' role described as a 'Zero LLM cost' agent that executes simple bash commands directly via
execSync. - [INDIRECT_PROMPT_INJECTION]: The skill's architecture involves 'Scout' agents reading the codebase to generate 'pheromones' (metadata) that influence the behavior of 'Worker' and 'Lead' agents. This creates a surface where malicious content in the codebase could influence the orchestration logic.
- Ingestion points: Project files and dependencies scanned by the Scout and Worker roles using Read, Grep, and Glob tools.
- Boundary markers: The skill uses specific role-based templates and instruction blocks to guide agent behavior, though it does not explicitly define delimiters for external data.
- Capability inventory: The system has the power to spawn new agents, send inputs to them, modify files (Write/Edit), and execute shell commands (Bash).
- Sanitization: The skill does not explicitly describe sanitization or validation of the 'pheromone' metadata collected from the codebase before it is used by other agents.
Audit Metadata