Log inSubmit server
Jacobian logo

Jacobian

Searches and executes typed mathematical operations for AI agents through MCP, with exact results and explicit uncertainty.

InstallRepository
Full README Compare alternatives Embed widget Claim this listing

Install Jacobian

Pick your client, then add it in one click or paste the config.

STDIONode.js
~/Library/Application Support/Claude/claude_desktop_config.json
claude_desktop_config.json
{
  "mcpServers": {
    "jacobian": {
      "command": "npx",
      "args": [
        "-y",
        "jacobian@latest"
      ]
    }
  }
}

Settings → Developer → Edit Config, merge this into mcpServers, then fully quit and reopen Claude Desktop.

Install not yet sandbox-tested— we haven't run this command through our automated sandbox yet. Not a red flag; we're working through the catalog.

Overview

Jacobian MCP server exposes a searchable vocabulary of typed mathematical operations through the `math.find` and `math.run` tools. Agents can inspect an operation's contract, execute it with structured input, and compose its typed result with later operations. Its library covers polynomial algebra, linear algebra, graph theory, SAT/SMT, finite algebra, probability, geometry, and topology. Reach for it when an agent needs bounded, independently executable mathematical steps rather than a prebuilt solver or workflow.

Use cases

  • Search for a typed operation matching a mathematical need
  • Execute exact rational matrix calculations
  • Compose graph, algebra, or geometry operations
  • Verify bounded SAT or SMT results
  • Inspect operation contracts before invocation

Key features

  • Searchable typed mathematical operation vocabulary
  • Contract inspection through MCP
  • Structured execution with typed results
  • Exact linear algebra and polynomial computation
  • Graph, SAT/SMT, algebra, probability, geometry, and topology operations
  • CLI and native Python access

Security & health

What it can reach and how reliably it runs — from our own checks, not the maintainer’s claims.

ActiveLast checked Today

Recent health check succeeded.

Supply-chain signal

No high-severity advisories surfaced by our automated scan.

Critical 0High 0Medium 0Low 0

Scanned 12d ago via OSV.dev · jacobian@latest (npm)

Adoption & maintenance

Factual signals from GitHub, npm, and our automated checks — not a rating.

GitHub stars318Stargazers on the source repository.
npm downloads722Package downloads in the last 30 days.
Last commit3d agoMost recent push to the default branch.
Directory activity1 viewsConfig copies, upvotes, and views on AllMCPs.

How Jacobian works

What the Jacobian MCP server does

The Jacobian MCP server gives AI agents access to narrowly scoped mathematical operations with typed inputs, outputs, and stated postconditions. An agent can search the operation vocabulary for a needed result, inspect the exact contract, and then execute the selected operation. The returned value is intended to be used as a typed input to another operation, allowing a reasoning model to compose calculations without requiring Jacobian to prescribe the overall proof or workflow.

The built-in portfolio includes polynomial maps and polynomial algebra, exact linear algebra, graph paths and colorings, graph isomorphism, bounded SAT and SMT solving, finite algebra, probability, geometry, and topology. Results are exact when the operation claims exactness; approximation, incompleteness, and uncertainty are represented explicitly where relevant.

How it works

The MCP interface provides two primary mathematical tools:

  • math.find searches for an operation by mathematical need or inspects one known operation's contract.
  • math.run invokes an operation with its structured payload and returns the typed result.

A typical agent workflow is to call math.find, inspect a suitable contract, and then call math.run once the input shape and expected result are clear. For example, an integer extended-gcd operation can accept two values and return the greatest common divisor with Bézout coefficients. A later operation can consume those typed values.

The same mathematical library is available outside MCP through the command-line interface and a native Python API. The Python distribution includes the kernel, CLI, MCP server, and maintained backend stack based on SymPy, NetworkX, Z3, and Python-FLINT.

Setup and configuration

The Jacobian MCP server can be launched without a global install using uvx --python 3.12 --from jacobian jacobian-mcp. Hosts that require an npm command can use npx jacobian mcp; the npm package maps its package version to the corresponding uvx invocation. A persistent installation is available with python -m pip install jacobian, followed by jacobian-mcp.

The setup helper is npx jacobian@latest setup. It detects supported agents, lets the user review changes, and writes registrations only after confirmation. It requires Node.js 20.17+, 22.13+, or 23.5+, plus uvx on PATH; it does not install Node.js, Python, uv, or an agent. Existing registrations can be refreshed with npx jacobian@latest upgrade.

The tested binary-install contract is CPython 3.12 or 3.13 on glibc Linux x86-64. Other systems may work if upstream wheels are available, but they are outside the tested release contract. Alpine/musl cannot install the complete mandatory stack from PyPI.

Tools and capabilities

The server's mathematical operations are designed as reusable primitives rather than large domain-specific solvers. Z3-backed SAT and SMT operations use the maintained Python binding directly. The operation library also supports exact computations in several algebraic, combinatorial, geometric, and topological domains. The CLI provides corresponding inspection and execution flows for local terminal use.

Limitations and notes

Jacobian is pre-stable, so experimental operation contracts may change between releases. Singular and QEPCAD are optional system runtimes needed by a small subset of operations; the Python installation does not provision them. Ordinary operations do not require Docker. Backend availability can be checked through the Python API or MCP inspection.

The Jacobian MCP server supplies mathematical moves, not a complete reasoning strategy. The calling model remains responsible for selecting operations, composing results, and deciding when its reasoning is complete. The project is licensed under MIT.

Read the full README View source on GitHub

Reviews

No reviews yet — be the first to share how it worked for you.

Frequently Asked Questions about Jacobian

Jacobian is an MCP server that connects AI agents to a searchable library of typed mathematical operations. Its main tools, `math.find` and `math.run`, let agents discover or inspect an operation, execute it with structured input, and compose the typed result with subsequent calculations.

Install it with `uvx --python 3.12 --from jacobian jacobian-mcp`, or use `python -m pip install jacobian` followed by `jacobian-mcp`. Hosts requiring npm can run `npx jacobian mcp`; the setup helper is `npx jacobian@latest setup`.

Jacobian provides `math.find` and `math.run` through MCP. Use `math.find` to search for a mathematical operation or inspect its contract, then use `math.run` to execute it with the required structured payload and receive its typed result.

Yes, Jacobian includes Codex among the agents that its setup command can configure. Run `npx jacobian@latest setup --codex` to preview or apply a Codex registration, using `--dry-run` to preview the plan and `--yes` for noninteractive setup.

Jacobian supports Python 3.12 and 3.13 under its tested binary-install contract on glibc Linux x86-64. Other systems may work when compatible upstream wheels exist, but Alpine/musl cannot install the complete mandatory stack from PyPI.

Related MCP servers

All in Data Science Tools All alternatives
Add a badge to your READMEShow it’s listed — live SVG, Markdown or HTML.

AllMCPs Directory Badge

Full Badge Customizer

Showcase your server listing on GitHub or your project documentation. Embed this dynamic SVG badge to highlight official listing status and live engagement.

Badge Style:
Live Dynamic SVG PreviewJacobian AllMCPs Directory Badge
Markdown (GitHub README)
[![AllMCPs](https://allmcps.com/api/badge/jacobian?style=directory)](https://allmcps.com/mcp/jacobian)
HTML Embed
<a href="https://allmcps.com/mcp/jacobian"><img src="https://allmcps.com/api/badge/jacobian?style=directory" alt="Jacobian on AllMCPs" /></a>