Skip to content

Brahe MCP

Brahe MCP is a Model Context Protocol server that exposes Brahe's astrodynamics capabilities to language models and AI-assisted development tools. Once connected, a model can call Brahe functions directly — time conversions, coordinate and frame transformations, orbit propagation, access computation, and TLE/ephemeris lookups — without the model having to write or run Brahe code manually. This can significantly improve the accuracy of astrodynamics-related responses and enable new use cases like interactive analysis and natural language querying of satellite data.

The server is published to PyPI as brahe-mcp. It runs locally over stdio and works with any MCP-compatible client, including Claude Desktop, Claude Code, Gemini CLI, and the OpenAI Codex CLI.

Installation

Install the MCP server with uv (recommended) or pip:

1
uv tool install brahe-mcp

1
pip install brahe-mcp

Both methods expose a brahe-mcp command on the PATH. MCP clients launch that command to start the server over stdio.

Client Configuration

The server entry is the same across clients — register a server named brahe that runs the brahe-mcp command. Only the config file location and syntax differ by client.

Claude Desktop

Edit the Claude Desktop config file:

  • macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
  • Windows: %APPDATA%\Claude\claude_desktop_config.json

Add a brahe entry under mcpServers:

1
2
3
4
5
6
7
{
  "mcpServers": {
    "brahe": {
      "command": "brahe-mcp"
    }
  }
}

Restart Claude Desktop for the change to take effect.

Claude Code

Claude Code reads MCP servers from .claude/settings.json (project-scoped) or from the user-level settings file. Use the same JSON block as Claude Desktop:

1
2
3
4
5
6
7
{
  "mcpServers": {
    "brahe": {
      "command": "brahe-mcp"
    }
  }
}

Alternatively, add the server via the CLI without editing JSON directly:

1
claude mcp add brahe brahe-mcp

OpenAI Codex CLI

Codex CLI stores MCP configuration in TOML at ~/.codex/config.toml (or project-scoped .codex/config.toml):

1
2
3
[mcp_servers.brahe]
command = "brahe-mcp"
args = []

Or add it via the Codex CLI:

1
codex mcp add brahe -- brahe-mcp

ChatGPT Desktop and Web

ChatGPT Desktop and the ChatGPT web client do not support local stdio MCP servers — they require remote HTTPS endpoints. Brahe MCP is currently a local-only server, so it cannot be connected to ChatGPT at this time. Use Claude Desktop, Claude Code, Codex CLI, or another MCP client with stdio support.

SpaceTrack Credentials

The SpaceTrack-backed tools (TLE lookups, conjunction data) require a Space-Track.org account. Provide credentials through the env block in the server config:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
{
  "mcpServers": {
    "brahe": {
      "command": "brahe-mcp",
      "env": {
        "SPACETRACK_USER": "your@email.com",
        "SPACETRACK_PASS": "your-password"
      }
    }
  }
}

For Codex, pass them through the CLI:

1
2
3
4
codex mcp add brahe \
  --env SPACETRACK_USER=your@email.com \
  --env SPACETRACK_PASS=your-password \
  -- brahe-mcp

Claude Code inherits the parent shell environment, so credentials already exported in ~/.zshrc or ~/.bashrc are available without an env block. Claude Desktop does not expand shell variables like ${SPACETRACK_USER} — literal values must be written in the config file.

Without these credentials, the CelesTrak-backed tools continue to function; only the SpaceTrack tools will return an authentication error.

Local Development

To run the server from a local clone of the repository (useful for development or testing unreleased features):

1
2
3
git clone https://github.com/duncaneddy/brahe-mcp.git
cd brahe-mcp
uv sync --group dev

Point your MCP client at the local clone with uv run:

1
2
3
4
5
6
7
8
{
  "mcpServers": {
    "brahe": {
      "command": "uv",
      "args": ["run", "--directory", "/absolute/path/to/brahe-mcp", "brahe-mcp"]
    }
  }
}

Replace /absolute/path/to/brahe-mcp with the full path to your clone.

Further Reading