lean4
Pass
Audited by Gen Agent Trust Hub on Apr 16, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill instructions and documentation focus entirely on legitimate Lean 4 development workflows, including formalization, debugging, and pedagogy.
- [EXTERNAL_DOWNLOADS]: The skill fetches data from files and URLs for formalization and search tasks. It utilizes well-known community tools and services such as GitHub, LeanSearch, and Loogle, which are documented neutrally and serve the skill's primary purpose.
- [COMMAND_EXECUTION]: The skill executes Lean-specific commands (lake, lean) and local utility scripts to manage projects and verify proofs. These operations are essential for theorem proving and are executed within the expected developer environment.
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external content (PDFs, papers, URL sources). It identifies this surface and implements safety mechanisms such as boundary markers and reliance on formal verification via the Lean type-checker to ensure that only valid mathematical proofs are accepted.
Audit Metadata