maintain-verification-skill
Pass
Audited by Gen Agent Trust Hub on Nov 6, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted project source code to generate verification instructions, creating a surface for indirect prompt injection. 1. Ingestion points: Feature source files are read during the 'Source wave' (Step 2). 2. Boundary markers: No explicit delimiters or instructions to ignore embedded commands are specified for the subagents. 3. Capability inventory: The coordinator can execute shell commands (harness scripts), drive the application, and write files within the skill's own directory. 4. Sanitization: No sanitization or validation of the generated recipes is mentioned before they are used in the 'Live pass'.
- [COMMAND_EXECUTION]: The skill manages and executes local harness scripts and facilitates automated application driving to verify features, which involves running executable code and potentially invasive automated tooling.
Audit Metadata