nimbus-mcp

nimbus-mcp

A launcher that exposes your local Nimbus index and agents to any MCP client by locating and execing the Nimbus CLI binary.

Category
Visit Server

README

@nimbus-dev/mcp

What this is

A tiny launcher that exposes your local Nimbus index and agents to any MCP (Model Context Protocol) client — editors, chat clients, or other MCP-speaking tools. It does no work itself: it locates the Nimbus CLI binary already installed on your machine and execs it as nimbus mcp-server --stdio, then gets out of the way and lets stdio pass straight through.

This package is MIT-licensed and does not depend on (or import from) the AGPL-3.0 packages/cli or packages/gateway Nimbus source — it only knows how to find the installed binary, never how to run the gateway itself.

Requires the Nimbus gateway to already be installed. This launcher does not install or bundle Nimbus; run the regular Nimbus installer first, and keep the gateway configured the way you normally use it (the mcp-server --stdio command talks to your existing local index).

Install

Not published to npm yet. @nimbus-dev/mcp has no release-please entry and no publish workflow — registry submission is deliberately deferred to a release activity. The npm / npx instructions below are what you will use once it is published; until then, use Running from a local checkout.

Run it directly, without a global install, from your MCP client's config (see below), or install it explicitly:

npm install -g @nimbus-dev/mcp

Quickstart

Add nimbus-mcp as a command in your MCP client's server configuration, for example:

{
  "mcpServers": {
    "nimbus": {
      "command": "npx",
      "args": ["-y", "@nimbus-dev/mcp"]
    }
  }
}

Running from a local checkout

Until the package is published, point your MCP client straight at the launcher's entry point in a clone of the Nimbus repository. Use absolute paths — an editor-spawned MCP server does not inherit your shell's working directory:

{
  "mcpServers": {
    "nimbus": {
      "command": "bun",
      "args": ["/absolute/path/to/nimbus-mcp/src/index.ts"]
    }
  }
}

If you already have the Nimbus CLI installed, you can skip the launcher altogether and have your client run nimbus mcp-server --stdio directly — the launcher exists only to find that binary for you.

How the binary is found

The launcher looks for the Nimbus binary in this order:

  1. NIMBUS_BIN — an explicit full path to the binary, if set. A NIMBUS_BIN that points at a non-existent file is reported as an error, never silently ignored.
  2. PATH — the first nimbus (or nimbus.exe on Windows) found on your PATH.
  3. Known per-platform install directories, the installer's own output directory first: %LOCALAPPDATA%\Programs\Nimbus\bin on Windows, ~/.local/bin on macOS and Linux, then /opt/homebrew/bin and /usr/local/bin (macOS) or /usr/local/bin and /usr/bin (Linux).

Step 3 is what carries a GUI-launched editor on macOS, which typically spawns MCP servers without your shell's PATH. If none of those resolve, the launcher exits with a message naming the fix — never a bare exit code.

Environment variables

  • NIMBUS_BIN — override the resolved binary path. Point it at the exact Nimbus CLI executable.
  • NIMBUS_MCP_TIMEOUT_MS — how long the Nimbus CLI's MCP adapter waits for an agent brief before returning a clean timeout error; it defaults to 60 s. Lower it when your editor's own MCP transport timeout is shorter, so the tool reports a timeout instead of having the call severed underneath it. This is the adapter's own budget — the gateway imposes no such timeout. It is read by the underlying nimbus mcp-server process, which this launcher execs with your environment inherited unchanged.

See also

  • docs/architecture.md — Nimbus subsystem design and the MCP connector standard.
  • The gateway's mcp-server command, in the main Nimbus repository (packages/gateway, packages/cli) — this package launches it, but does not contain or license it.

License

MIT — see LICENSE. Note that the Nimbus gateway and CLI this launcher execs are licensed separately under AGPL-3.0.

Recommended Servers

playwright-mcp

playwright-mcp

A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.

Official
Featured
TypeScript
Magic Component Platform (MCP)

Magic Component Platform (MCP)

An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.

Official
Featured
Local
TypeScript
Audiense Insights MCP Server

Audiense Insights MCP Server

Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.

Official
Featured
Local
TypeScript
VeyraX MCP

VeyraX MCP

Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.

Official
Featured
Local
graphlit-mcp-server

graphlit-mcp-server

The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.

Official
Featured
TypeScript
Kagi MCP Server

Kagi MCP Server

An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.

Official
Featured
Python
E2B

E2B

Using MCP to run code via e2b.

Official
Featured
Neon Database

Neon Database

MCP server for interacting with Neon Management API and databases

Official
Featured
Exa Search

Exa Search

A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.

Official
Featured
Qdrant Server

Qdrant Server

This repository is an example of how to create a MCP server for Qdrant, a vector search engine.

Official
Featured