skills/trancee/skills/lean/Gen Agent Trust Hub

lean

Pass

Audited by Gen Agent Trust Hub on Oct 18, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPRIVILEGE_ESCALATIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill facilitates the installation of Lean 4 using official scripts from elan.lean-lang.org and identifies github.com/leanprover-community/mathlib4.git as a trusted source for mathematical libraries.
  • [COMMAND_EXECUTION]: A helper script, scripts/check-axioms.py, executes local build processes using the lake tool. It uses subprocess.run with list-based arguments and enforces strict regex validation ([A-Za-z_][A-Za-z0-9_']*(?:\.[A-Za-z_][A-Za-z0-9_']*)*) on module and theorem names to prevent shell injection.
  • [PRIVILEGE_ESCALATION]: Setup instructions for Windows environments suggest a temporary bypass of the PowerShell execution policy to run the elan installer, which is a standard procedure for this ecosystem and is appropriately scoped to the installation task.
  • [INDIRECT_PROMPT_INJECTION]: The skill acknowledges that Lean project metadata and tactics can contain executable logic. It provides actionable security guidance, such as reviewing code before building and using sandboxing tools like bubblewrap to process untrusted proof submissions.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 18, 2026, 02:24 PM