Numina Lean Agent — Theorem Proving Toolkit

Trust Score 76/100

A Lean 4 theorem-proving assistant: search lemmas, verify proofs, repair and simplify proofs, and get LLM-assisted informal proofs.

triggers:leantheorem provingverify proofsearch lemmarepair proofinformal prover

GitHub SKILL.md

What it does

Numina Lean Agent provides a suite of skills for working with Lean 4 theorem proving workflows. It exposes search tools for finding lemmas, verification tools to check and disprove proofs, code-transform utilities to repair or simplify proofs and extract theorems, and LLM-powered helper tools for informal proof generation and discussion.

When to use it

When searching for relevant lemmas or premises during proof development.
When you need to verify, check, or attempt to disprove a Lean proof.
When refactoring, repairing, or simplifying theorem code.
When you want LLM-assisted informal proofs, explanations, or code-golf-style challenges for Lean code.

What's included

Scripts: repository provides structured skill directories (search, verification, code-transform, llm) — individual SKILL.md files per subskill.
References: subfolders likely contain implementation tools and CLIs for interacting with Lean and external search providers.
Instructions: the top-level SKILL.md indexes subskills and lists required API environment variables (GEMINI_API_KEY, OPENAI_API_KEY, ANTHROPIC_API_KEY, AXLE_API_KEY).

Compatible agents

Agents with code execution and LLM integration (OpenAI, Anthropic, Gemini-capable agents), plus tooling for Lean integration.

Audit Summary

Numina Lean Agent is a Lean 4 theorem-proving toolkit that serves as an index to four sub-skills (search, verification, code-transform, llm). The main SKILL.md is minimal — just a table of sub-skills and required environment variables. No scripts were bundled for testing. The skill references standard API keys (Gemini, OpenAI, Anthropic, Axle) for external services, which is normal. No security concerns detected.

Watch Out

Requires GEMINI_API_KEY, OPENAI_API_KEY, ANTHROPIC_API_KEY, and/or AXLE_API_KEY for full functionality
Main SKILL.md is just an index — actual logic lives in sub-skill files not included in the audit

Notes

Lightweight index-only skill. Would benefit from more context in the main SKILL.md about how the sub-skills compose together.

Information

Repository: numina-lean-agent
Stars: 242

Trust Score

Overall76

Security95

Code Quality55

Architecture58

Usefulness42

Related Skills

Retriever (Membrane CLI)

Integrate Retriever search and discovery via the Membrane CLI — manage connections and run queries across Retriever data sources.

Retriever (Membrane) Integration

Integrate with Retriever via the Membrane CLI to search, query, and proxy Retriever data (documents, snippets, queries) with built-in auth and pagination.

ClawPod / Massive Unblocker

Bypass anti-bot restrictions and fetch rendered HTML or structured search results via Massive's Unblocker API (handles CAPTCHAs, JS rendering, geo-restrictions)

Algolia Cost Tuning

Practical guide to reduce Algolia billing by auditing records/requests, switching to virtual replicas, multi-query usage, caching, and cleanup strategies.

Memory Search

Search past coding sessions and observations using natural-language queries to retrieve decisions, notes, and context.

Research Information Lookup

Find and synthesize current research and academic papers with automatic backend routing (Parallel Chat API or Perplexity) and structured citations.

Heptabase CLI — Terminal access to Heptabase MCP

Command-line interface for interacting with a Heptabase knowledge base: search notes, read cards/journals, append entries, and search/read PDFs — useful for ter

Voice Memo Organizer

Locate, transcribe (local whisper.cpp), summarize and index Apple Voice Memos into a searchable archive with titles, themes and key quotes.

Back to Skills