create-formal-verification-skill
Pass
Audited by Gen Agent Trust Hub on Oct 26, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill analyzes the codebase to identify verification targets and properties, which exposes it to instructions or malicious patterns embedded in the code.
- Ingestion points: Repository codebase (e.g., 'Answer from the codebase', 'Choose the target from the code').
- Boundary markers: None present to delimit codebase content from the generation process.
- Capability inventory: The skill involves file creation ('Generate the verify skill') and command execution ('run the check', 'Follow the generated skill').
- Sanitization: No sanitization of the input codebase is specified.
- [DYNAMIC_EXECUTION]: The skill requires the agent to generate new project-local skills and verification models (scripts) at runtime and subsequently execute them to confirm they catch bug variants.
Audit Metadata