maintain-verification-skill
Pass
Audited by Gen Agent Trust Hub on Oct 17, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill identifies and processes project-local source files and feature maps to detect "doc drift." Untrusted content within these files could potentially influence the agent's summaries, verification recipes, or the content of the resulting pull request.
- Ingestion points: Project source code and feature definition files (READMEs and siblings) typically located in
.cursor/skills/verify-*/. - Boundary markers: The instructions lack explicit delimiters or warnings to ignore instructions embedded within the source files being audited.
- Capability inventory: The skill can execute local harness scripts, drive application instances (CLI, Server, or UI), and submit pull requests with modifications to the skill directory.
- Sanitization: No sanitization, escaping, or validation of the content read from source files is performed before interpolation into summaries or recipes.
- [COMMAND_EXECUTION]: As part of its primary function, the skill executes local "harness scripts" and drives application processes (CLIs, servers, or UIs) according to the launch model defined in the target verification skill. This involves the execution of code residing within the project's own directory structure.
Audit Metadata