MCP & API Keys

The Model Context Protocol (MCP) is an open standard that lets AI assistants connect to external data sources. Gatlas provides an MCP server that gives AI tools like Claude, Cursor, and others direct access to query your synced portfolio data, check sync health, and browse available schemas — all scoped to exactly what you want to share.

• Organization — Full access to all entities and views. Use for trusted internal users who need to see the full portfolio.
• Hierarchy — Access to a specific entity and all its children. Good for portfolio managers who oversee a sub-group of companies.
• Explicit — Access to specific entities and/or views only. Most restrictive — use for external operators or limited-scope analysts.

When you create a key, Gatlas generates a setup link you can share. The recipient visits the link and sees:

• Their API key (shown once)
• Setup instructions for Claude Code, Claude Desktop, and Cursor
• Example prompts they can try
• What data they have access to

If you provide a recipient email, Gatlas sends the setup link automatically. You can also copy the link and share it however you prefer. For a guide written for the recipient, see Using MCP Access.

• list_entities — List portfolio companies
• list_tables — Browse available database tables
• query_sql — Run SQL queries against synced data
• list_views — List published views
• describe_view — Get view details, schema, and AI context
• query_view — Query a published view with filters
• get_sync_status — Check connection sync health
• get_recent_syncs — Recent sync run history
• trigger_sync — Manually trigger a data sync