Connect your AI assistant to geoDB
geoDB has a connector for AI assistants: the remote MCP server at https://api.geodb.io/mcp (Streamable HTTP). The person signs in to geoDB from inside their assistant (OAuth) and ticks the projects it may reach; the assistant then works with their own permissions, every call shows in their access log (in geoDB: Settings → Connected AIs & keys → Activity), and they can disconnect at any time. No key to copy.
https://api.geodb.io/mcp
Steps for your assistant
Claude (claude.ai, Claude Desktop, Cowork, the mobile apps)
- Free, Pro or Max: open Customize → Connectors, choose "+ Add" → "Add custom connector", name it geoDB, paste https://api.geodb.io/mcp, then continue and Add.
- Team or Enterprise: an owner adds it once under Organization settings → Connectors → Add → Custom → Web (same URL); each member then finds it under Customize → Connectors and clicks Connect.
- Sign in to geoDB when asked and tick the projects the assistant may reach.
- So Claude's own code can reach geoDB (big tables, photos, files): on a Free, Pro or Max plan, open Settings → Capabilities and turn on "Allow network egress" (under "Code execution and file creation"). On a Team or Enterprise plan only an owner can change it: Organization settings → Capabilities → "Allow network egress to package managers and specific domains", then add api.geodb.io and geodbstorage.blob.core.windows.net (or choose "All domains"); Team plans start at package managers only and Enterprise with it off.
A Free plan holds one custom connector.
Claude Code
- Run: claude mcp add --transport http geodb https://api.geodb.io/mcp
- Then /mcp inside Claude Code (or claude mcp login geodb) and sign in to geoDB in the browser.
ChatGPT
- Go to chatgpt.com/plugins, select the plus button, then "Add custom MCP server"; give it a name (geoDB) and a description.
- Under Connection, enter https://api.geodb.io/mcp; configure authentication (geoDB signs you in with OAuth), review the risk warning and select "I understand and want to continue", then "Create as a plugin".
- Sign in to geoDB, tick the projects, then pick it in a chat (type @).
Your account's or workspace's policies decide whether custom MCP servers are allowed; OpenAI's docs do not list plans for it, so if the option is missing, check with your workspace admin or OpenAI's help. ChatGPT's data-analysis Python environment cannot make web requests or call APIs, so it cannot use a session_key or the Python client: ChatGPT works through the connector's own reads and writes. For a whole table, the connector's export_link gives you a download link: download the file and upload it to the chat.
Grok
- Go to grok.com/connectors, click New Connector, select Custom, enter https://api.geodb.io/mcp and sign in to geoDB.
- On Grok Business or Enterprise a team admin provisions it first, in the console at console.x.ai: on the connectors page, "+ Add Connector" → Other → the URL; members then connect it at grok.com/connectors.
Any other MCP client
- Add a remote MCP server at https://api.geodb.io/mcp with the Streamable HTTP transport.
- It authenticates with OAuth 2.1 (dynamic client registration or a client ID metadata document); the protected-resource metadata is at /.well-known/oauth-protected-resource/mcp on the same host.
Assistants rename their menus often: if a name above differs, follow the assistant's own current help for "custom connector" or "remote MCP server".
Once connected
Once connected, ask your assistant "What can you see in geoDB?". It reads geoDB's own instructions first, then works through the projects you ticked. For heavy work in its own code sandbox it asks the connector for a session_key (one hour), installs the client with pip install "geodb-client>=0.3.4,<0.4" and runs the two-line self-test the key comes with. Every write is a dry run first, is sent only after your yes, and can be undone.
Software and vendors: the open protocol
Software and vendors without an assistant read geoDB with a key a project owner issues (Project settings → API access grants) — see AGENTS.md in the geodb-protocol repository.
- The protocol repository (open spec, JSON Schemas, examples, conformance runner)
- AGENTS.md: the entry point written for AI agents
- API reference and the OpenAPI schema
- llms.txt: the index of what an AI model should read
Questions: contact geoDB.