cookbook-migrate-model
Pass
Audited by Gen Agent Trust Hub on Oct 2, 2026
Risk Level: SAFECOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: Employs standard developer tools including
git,gh(GitHub CLI), andgrepfor repository management, such as checking out specific PRs or commits and identifying reference anchors. It also usesmint(Mintlify CLI) for documentation validation and local previewing.\n- [DYNAMIC_EXECUTION]: Recommends a local development workflow where legacy logic is ported into temporary Node.js scripts for code generation and whereindirect evalis used in throwaway audit scripts to verify that migrated command outputs match legacy versions.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes data from existing repository files and external maintainer-provided inventory notes.\n - Ingestion points: Legacy cookbook
.jsxand.mdxfiles and maintainer inventory notes provided via prompt instructions.\n - Boundary markers: Strictly defined 'Hard rules' (e.g., 'Never modernize', 'Never invent versions') and a 'Verified policy' act as constraints to ensure faithful transcription of data.\n
- Capability inventory: Includes local file system access, version control operations, and local script execution via
node.\n - Sanitization: Achieved through a mandatory equivalence audit and a requirement for reproducible version anchors like commit hashes.
Audit Metadata