ps-maintain-verification-skill
Pass
Audited by Gen Agent Trust Hub on Sep 17, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from the project source code and external skill files, which can influence agent behavior.
- Ingestion points: Project source code files (Step 2: Source wave) and target verification skill files located in
.cursor/skills/verify-*/(Step 0, 1). - Boundary markers: Absent. The instructions do not specify delimiters or warnings to ignore instructions embedded within the source code or feature maps.
- Capability inventory: The skill can launch subagents, write to local files in the skill directory, and execute shell commands via harness scripts.
- Sanitization: Absent. There is no mention of filtering or validating the content read from source files before it is used to generate summaries or drive the application.
- [COMMAND_EXECUTION]: The skill is designed to execute shell commands as part of its 'Live pass'.
- Evidence: Step 4 ('Live pass') involves executing the verification skill's own launch model and harness scripts to exercise features.
- [DYNAMIC_EXECUTION]: The skill identifies and executes code at runtime based on the contents of target directories.
- Evidence: It dynamically locates target skills in
.cursor/skills/verify-*/and executes their associated 'harness scripts' and 'launch sections' (Step 0 and Step 4).
Audit Metadata