MCP servers that ship as code — a repository or a package you install and run locally (usually over stdio). They have no hosted endpoint, so we cannot probe or grade them: instead each one shows its repository, install command, license, stars and when it was last updated.
Collected from GitHub repositories and npm, PyPI, Cargo, Go, Maven, NuGet and RubyGems packages tagged as MCP servers, the official MCP Registry, Docker's MCP catalog, ModelScope, Baidu MCP World, Cline and other public directories; one entry per repository. A server that also offers a hosted endpoint is listed under Remote instead. These entries are not counted in our reachability, verification or readiness statistics.
-
lean-local-search-mcpgithub.com/dududuguo/lean-local-search-mcpupdated this year
MCP server for fast local Lean and Mathlib theorem search with tree-sitter indexing and proof probes
MIT ★ 1 -
Lean theorem prover MCP: automated proof search, mathlib integration, job management
mcpb MIT ★ 3 -
Airtight math tools an AI uses over MCP — 3.7M-theorem search, PSLQ constant ID, OEIS, real Lean kernel checks, applicability checklists. No LLM inside, no API key.
pypi Apache-2.0 ★ 13 -
High-performance WebAssembly theorem prover with dependent types, hash-consing (150x faster), Ed25519 proof signatures, MCP support for Claude Code, AgentDB vector search, episodic memory, and ReasoningBank learning. Formal verification with cryptographic
npm Apache-2.0 -
TheoremSearch-MCPgithub.com/FDscend/TheoremSearch-MCPupdated this year
MCP wrapper for TheoremSearch (UW Math AI Lab) — semantic search over 9.27M mathematical statements across arXiv, Stacks Project, ProofWiki & more. Includes official remote MCP registration plus a local Python server with 5 tools (filters + dependency graph).
MIT -
dsh-lean-mcpactive
Bridges the Lean 4 language server into DeepSeek Harness so agents can inspect proof goals, run Lean code, build projects, and search Mathlib.
npm MIT -
Between having a mathematical idea and having a proof of it there is a lot of work that is not proving: find the object, break the false claims, measure what survives, reduce it to what it really is, assemble the rest. certo does that work, and every step returns a certificate anyone can re-check without trusting it. CLI + MCP.
pypi MIT ★ 1