maintain-verification-skill
Pass
Audited by Gen Agent Trust Hub on Sep 1, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes project source code and feature maps to audit verification logic. This creates a vulnerability surface where data-resident instructions could influence the agent's actions during the analysis phase.
- Ingestion points: Reads feature files, feature maps, and product source code (Step 2 and Step 3).
- Boundary markers: No explicit separators or instructions to ignore embedded content are provided.
- Capability inventory: The skill has the authority to execute shell commands, manage processes, and submit Pull Requests (Step 4, Step 5, and Step 6).
- Sanitization: No sanitization or validation of the ingested source data is described.
- [COMMAND_EXECUTION]: The skill's core functionality requires driving the application live and executing local harness scripts.
- Evidence: The 'Live pass' section instructs the agent to health-check processes, relaunch instances, and exercise features. It also explicitly allows fixing and executing harness scripts provided they are documented and executable.
Audit Metadata