Vibe Math MCP

Vibe Math MCP is an MCP server that brings high-performance mathematical computation to LLMs using Python scientific libraries. Built on Polars, Pandas, NumPy, SciPy, and SymPy, the server connects AI assistants directly to computational engines to handle complex operations without rounding errors or arithmetic hallucinations. Data scientists, financial analysts, software developers, and researchers use this tool to execute symbolic and numerical tasks within their agent workflows. It provides twenty-one specialized tools organized across foundational calculations, array manipulation, statistics, financial modeling, linear algebra, and calculus. The server also includes a batch execution tool that chains dependent calculations into a single call, reducing token consumption. Every tool supports customizable output verbosity modes ranging from complete metadata payloads to minimal or value-only formats, helping agents conserve context space while retrieving accurate mathematical answers.

Category: Data & Analytics

Tags: mathematics, numpy, pandas, polars, scipy, sympy

Visit Vibe Math MCP

How to install and configure Vibe Math MCP

1. Ensure you have uv installed on your system. 2. For Claude Desktop, open Settings > Developer > Edit Config and add the following server configuration: json { "mcpServers": { "Math": { "command": "uvx", "args": ["vibe-math-mcp"] } } } 3. For Claude Code, run the command: bash claude mcp add --transport stdio math -- uvx vibe-math-mcp 4. For Cursor or VS Code, use the integrated installation link or configure a project-level .mcp.json file using the same uvx vibe-math-mcp execution command. 5. Restart your client and verify using claude mcp list or /mcp.

What you can do with Vibe Math MCP

• Solving systems of linear equations or computing matrix decompositions such as SVD, QR, and Cholesky directly inside LLM chat sessions. - Calculating present and future value, internal rate of return, and compound interest across various payment frequencies for financial modeling. - Performing symbolic and numerical calculus, including computing definite integrals, derivatives, limits, and series expansions without mathematical hallucinations. - Analyzing datasets with descriptive statistics, correlation matrices, pivot tables, and array transformations like normalization and standardization. - Orchestrating multi-step mathematical workflows with batch execution to chain intermediate outputs while reducing prompt token overhead.

Key facts

• Open Source
• https://github.com/apetta/vibe-math-mcp
• Data & Analytics, Developer Tools & Code Intelligence
• mathematics, numpy, pandas, polars, scipy, sympy

Part of MCP Servers

How do I install Vibe Math MCP?

You can install Vibe Math MCP using uvx. In Claude Desktop, edit your config JSON to run command uvx with the vibe-math-mcp argument under mcpServers. For Claude Code, run the command claude mcp add --transport stdio math -- uvx vibe-math-mcp in your terminal.

What can Vibe Math MCP do?

Vibe Math MCP provides 21 mathematical tools covering basic calculations, array operations, statistics, financial calculations, linear algebra, and calculus. It leverages NumPy, SciPy, SymPy, Polars, and Pandas to handle matrix operations, equation solving, integrations, time value of money calculations, and multi-step batch executions.

Which MCP clients work with Vibe Math MCP?

Vibe Math MCP works with any client supporting the standard Model Context Protocol over stdio or HTTP. Verified clients include Claude Desktop, Claude Code, Cursor, and VS Code. It can also be run locally via HTTP server mode for custom container workflows.

How does Vibe Math MCP optimize token usage?

Every tool provides an output_mode parameter allowing clients to request full, compact, minimal, value-only, or final output formats. In addition, its batch_execute tool allows chaining multiple dependent operations in a single request, reducing context token consumption by up to 95 percent.

Is Vibe Math MCP open source?

Yes, Vibe Math MCP is open source software distributed under the MIT License. The code and test suite are publicly available on GitHub.