create-verification-skill
Pass
Audited by Gen Agent Trust Hub on Nov 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill analyzes project-local files to generate and then execute verification logic, creating a surface for indirect instructions to influence agent actions.
- Ingestion points: Reads repository files such as README.md, package.json, Makefiles, and source code (SKILL.md, Section 1).
- Boundary markers: Lacks explicit instructions for the agent to use delimiters or 'ignore' guards when processing potentially untrusted repository content.
- Capability inventory: The agent is authorized to execute generated shell commands, build binaries, install dependencies, and perform cleanup operations (SKILL.md, Section 2 & 4).
- Sanitization: No sanitization or verification of repo-derived strings (like script names or ports) is performed before they are used in shell executions.
- [COMMAND_EXECUTION]: The skill facilitates the automated execution of build and test commands (e.g., npm start, make dev, go run) found within the user's repository as part of the verification cycle.
Audit Metadata