Skip to content
dinobase

Claude Code

Dinobase works with Claude Code out of the box. Claude Code can use either the CLI (more token-efficient) or the MCP server — both give access to the same data and query engine.

Install

Section titled “Install”
Terminal window
curl -fsSL https://dinobase.ai/install.sh | bash -s -- claude-code

Installs Dinobase via uv, runs dinobase init, and writes CLI usage instructions to ~/.claude/CLAUDE.md. Then connect your data sources:

Terminal window
dinobase add stripe --api-key sk_test_...
dinobase add hubspot --api-key pat-...
dinobase sync

See Connecting Sources for the full list of 100+ supported sources, and Syncing & Scheduling for background sync options.

Claude Code can run shell commands directly, so no MCP configuration is needed. Just ask questions about your data:

“Query my Dinobase for customers who churned last quarter”

Claude Code will run commands like dinobase info, dinobase describe stripe.customers, and dinobase query "SELECT ..." to answer your question.

The CLI outputs JSON by default, which Claude Code parses automatically. This is 27% more token-efficient than MCP.

To re-run the setup step: dinobase install claude-code (safe to run multiple times — replaces the existing block).

Alternative: MCP server

Section titled “Alternative: MCP server”

If you prefer MCP, you can manually configure it. Run dinobase mcp-config claude-code to get the JSON config, then add it to your project’s .mcp.json.

Available commands

Section titled “Available commands”
CommandDescription
dinobase infoDatabase overview — sources, tables, row counts
dinobase statusSource freshness and sync history
dinobase describe <table>Table schema, column types, and sample data
dinobase query "<sql>"Execute SQL queries (DuckDB dialect)
dinobase refresh [source]Re-sync a data source
dinobase confirm <id>Execute a pending mutation
dinobase cancel <id>Cancel a pending mutation

Next steps

Section titled “Next steps”