This guide explains how to configure Goose AI agent framework with the Mapbox MCP Server.
- Goose installed on your system
- A Mapbox access token (get one here)
Follow the official Goose installation instructions.
Add the Mapbox MCP Server to Goose's configuration file (typically ~/.config/goose/profiles.yaml):
default:
provider: openai
processor: gpt-4
accelerator: gpt-4o-mini
moderator: passive
toolkits:
- name: developer
- name: mcp
requires:
servers:
- name: mapbox
command: npx
args:
- '-y'
- '@mapbox/mcp-server'
env:
MAPBOX_ACCESS_TOKEN: 'your_token_here'Alternatively, if you've built the server locally:
default:
provider: openai
processor: gpt-4
accelerator: gpt-4o-mini
moderator: passive
toolkits:
- name: developer
- name: mcp
requires:
servers:
- name: mapbox
command: node
args:
- '/path/to/mcp-server/dist/esm/index.js'
env:
MAPBOX_ACCESS_TOKEN: 'your_token_here'goose session startAsk Goose to create a map:
Show me a map of the Golden Gate Bridge
Goose supports MCP-UI, which means you'll see inline interactive maps rendered directly in the chat when you use the static_map_image_tool. This provides a richer visual experience compared to clients that only display static images.
For more information about MCP-UI features, see the MCP-UI documentation.
Verify that:
- Your Mapbox access token is correct
- Goose is properly configured (check
~/.config/goose/profiles.yaml) - The MCP server is accessible (check Goose logs)
Add the --disable-mcp-ui flag to args:
args:
- '-y'
- '@mapbox/mcp-server'
- '--disable-mcp-ui'For more information about Goose, visit the official Goose repository.