OpenMath Lean Theorem
Instructions
Set up the Lean proving environment, validate toolchains, and prove downloaded OpenMath theorems locally. Assumes the theorem workspace was already created by the openmath-open-theorem skill.
This skill does not run benchmark providers, prompt-based agent comparisons, or trace persistence workflows. Those belong to the separate openmath-lean-benchmark skill.
Workflow checklist
- Environment: Verify
lean,lake, andelanare installed and match the workspacelean-toolchain. - External skills: Install required Lean proof skills from leanprover/skills. Preferred manual install:
If you use preflight auto-install, first review the upstream repo and then pass an explicit target such asnpx leanprover-skills install lean-proof npx leanprover-skills install mathlib-build--install-dir .codex/skillsor--install-dir .claude/skillsso the write location is deliberate. Do not run auto-install without an explicit install dir. - Preflight: Run
python3 scripts/check_theorem_env.py <workspace>(see references/preflight.md). - Prove: Use
lean-proof/mathlib-buildskills to complete the proof. See references/proof_playbook.md for the OpenMath-specific proving loop. - Verify: Confirm
lake build -q --log-level=infopasses and nosorryremains. - Submit: Use the
openmath-submit-theoremskill to hash and submit the proof.
Scripts
| Script | Command | Use when |
|---|---|---|
| Preflight check | python3 scripts/check_theorem_env.py <workspace> |
After download, before proving; validates toolchain, required skills, and initial build. |
| Preflight (auto) | python3 scripts/check_theorem_env.py <workspace> --auto-install-skills --install-dir <path> |
Auto-install missing Lean skills during preflight into an explicit skills dir. |
Notes
- Lean version: Scaffolds pin
leanprover/lean4:v4.28.0andmathlib4 v4.28.0(set byopenmath-open-theorem'sdownload_theorem.py). - External skills: Not bundled. Required:
lean-proof,mathlib-build. Optional:lean-mwe,lean-bisect,nightly-testing,mathlib-review,lean-setup. Manualnpx leanprover-skills install ...is preferred; preflight auto-install additionally requiresgit, explicit user approval, and an explicit install dir. - Benchmarking: For agent evaluation, prompt comparison, or regression testing on the bundled Lean benchmark corpus, use the separate
openmath-lean-benchmarkskill.
References
Load when needed (one level from this file):
- references/preflight.md — Lean preflight command, skill checks, and initial build loop.
- references/proof_playbook.md — Step-by-step workflow for proving a downloaded Lean theorem locally.
- references/languages.md — Lean version and standard library.
