maintain-formal-verification-skill
Pass
Audited by Gen Agent Trust Hub on Oct 23, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill locates and executes shell commands defined within other local skill files (found in
.agents/skills/or.claude/skills/). This dynamic execution of commands is a core feature for running various project-specific verifiers. - [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data, specifically git commit messages and diffs, to determine required updates for verification models. This creates a surface where adversarial instructions in a commit message could potentially influence agent behavior.
- Ingestion points: Content from local skill files, git commit messages, and git diffs.
- Boundary markers: The instructions do not specify the use of delimiters or 'ignore' instructions when processing external data.
- Capability inventory: Includes shell command execution and file system write access for updating verification scripts and models.
- Sanitization: No explicit sanitization or validation of the commit messages or retrieved command strings is performed before they are used in the workflow.
Audit Metadata