executing-plans
Pass
Audited by Gen Agent Trust Hub on Sep 23, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill facilitates the execution of implementation plans stored in external files. This creates an attack surface where a plan file could contain malicious instructions intended to manipulate the agent.
- Ingestion points: Step 1 involves reading a plan file from the workspace.
- Boundary markers: The skill does not define specific delimiters for plan content, but it mandates a 'Review critically' step to identify concerns before execution.
- Capability inventory: The skill triggers task execution which includes file writes and verification commands (subprocess execution).
- Sanitization: The process requires identifying questions, raising concerns with a human partner, and stopping immediately if instructions are unclear or blockers are encountered.
- [COMMAND_EXECUTION]: The workflow requires the agent to 'Run verifications as specified' and 'Follow each step exactly,' which inherently involves executing shell commands or scripts defined within the implementation plan.
Audit Metadata