write-plans
Pass
Audited by Gen Agent Trust Hub on Oct 31, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill directs the agent to base plans on repository evidence, creating a vulnerability surface where adversarial instructions within the codebase could influence the agent's generated plans or execution steps.
- Ingestion points: Existing repository evidence and files (SKILL.md).
- Boundary markers: Not specified.
- Capability inventory: Creation, modification, and deletion of markdown files within 'docs/plans/' (SKILL.md).
- Sanitization: No sanitization of the ingested content is described.
- [COMMAND_EXECUTION]: The instructions involve verifying plan tasks through command execution and test results, which relies on the agent's shell capabilities.
- [DATA_EXFILTRATION]: No network access or automated data exfiltration patterns were identified in the instructions.
Audit Metadata