plan-writing
Pass
Audited by Gen Agent Trust Hub on Nov 1, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to analyze project-specific data to create execution plans, which creates a surface for indirect prompt injection. Malicious content within the project files could potentially manipulate the generated tasks.
- Ingestion points: Project files accessed through Read, Glob, and Grep tools.
- Boundary markers: There are no explicit delimiters or instructions provided to ignore or isolate instructions found within the project data during the planning phase.
- Capability inventory: The skill promotes tasks involving shell command execution (e.g., npm, curl) and the execution of local scripts (e.g., security_scan.py).
- Sanitization: No sanitization or validation of the ingested project data is performed before it is interpolated into the planning framework.
- [DYNAMIC_EXECUTION]: The skill encourages the agent to identify and incorporate the execution of project-local Python scripts, such as validators and audit tools, into the task plans based on the project type.
Audit Metadata