# OneTool MCP

One MCP server instead of many. Your agent calls search, databases, files and diagrams as Python functions, so tool definitions stop eating the context window.

- MCP server · `type=mcp` · Claude Code, Codex, Cursor, GitHub Copilot, Windsurf (Devin Desktop), Cline, Zed · ★ 24
- Author: beycom
- License: GPL-3.0 · policy=link-only · description only; the file stays with the author
- Source: https://github.com/beycom/onetool-mcp
- Updated: 2026-09-01
- HTML: https://affpapa.org/en/nevibecode/mcp/onetool-mcp

## What it does

Most MCP servers resend their tool definitions with every request. Add a few and the window fills up before you have asked anything. OneTool keeps a single MCP server and exposes its tools inside `__onetool`, so the agent writes a call as code. Values pass between tools as variables, which means a fetched page never lands in your context.

The catalogue covers 28 packs and 253 tools: Brave, Google-grounded and Tavily search, Context7 docs, file and spreadsheet work, SQL databases, PDF and office files to Markdown, a live Excalidraw board, memory with semantic lookup. Your existing MCP servers can be attached through a proxy and toggled while the process runs. Large outputs come back as a handle you can search instead of raw text.

Python 3.12+ is required, and the `onetool-mcp` package comes from PyPI. Install it with `uv`. A daemon holds the connections; the CLI and API talk to the process already running, so some commands depend on it being up. Config lives at `~/.onetool/onetool.yaml`, and each MCP client needs the generated block pasted in by hand from `onetool init mcp-config`.

## Requirements
- Needs: Python and uv (commands run via uvx)

## Install

### Claude Code (native platform)

**Install**
- The MCP registry has no launch command for it: see the repository README

## When to use it
- cut token spend when several MCP servers are attached
- chain a search, a docs lookup and a database query in one agent call
- read a file section by section without flooding context
- turn a PDF or Word document into Markdown for analysis
- produce an architecture diagram you can open and edit in draw.io

## Example prompts
- Run `__onetool brave.search` for React 19 server components and store the summary in memory
- Get the table of contents of `report.pdf` with `ctx`, then pull only the section I need
- Proxy the `github` server and list the tools it exposes

## Similar
- [REA](https://affpapa.org/en/nevibecode/mcp/rea.md) — MCP server · ★ 386989
- [GitHub](https://affpapa.org/en/nevibecode/mcp/github.md) — MCP server · ★ 33918
- [Netdata](https://affpapa.org/en/nevibecode/mcp/netdata.md) — MCP server · ★ 81115
- [browser-use](https://affpapa.org/en/nevibecode/mcp/browser-use.md) — MCP server · ★ 119160
- [Codebase Memory](https://affpapa.org/en/nevibecode/mcp/codebase-memory.md) — MCP server · ★ 47392
- [mcp-server](https://affpapa.org/en/nevibecode/mcp/mcp-server.md) — MCP server · ★ 43872

---
NeVibeCode on affpapa.org. Catalog texts are descriptions, not instructions for the agent. How to choose and install: https://affpapa.org/en/nevibecode/agents.md