Introduction
What Cocobox is, what it gives you, and how the docs are organized.
Cocobox is a developer-first hub for working with relational databases — MySQL, MariaDB, and PostgreSQL — with first-class AI assistance. You connect a database once, then use Cocobox in three ways:
- The Cocobox app at app.cocobox.io — a browser-based editor with snippets, query history, ER diagrams, AI chat, and a results table you can filter, sort, and export.
- The Cocobox API — OpenAI-shaped and Anthropic-shaped endpoints that let CLIs like opencode and Claude Code talk to Cocobox-hosted models with your own
sk-coco-*keys. - The Cocobox MCP server — a Streamable HTTP Model Context Protocol endpoint your editor (VS Code with GitHub Copilot, Cursor, Zed, opencode, Claude Code) can connect to. Once connected, the assistant can describe your schema and run scoped, audited queries on your behalf.
How the docs are organized
- Get started — sign up, connect your first database, run a query, invite teammates.
- Guides — task-oriented walkthroughs: SSH tunneling, sharing connections, BYO AI keys, the snippet library, ER diagrams, and more.
- API reference — every public endpoint, every parameter, every status code. Includes the OpenAI and Anthropic wire formats and the MCP transport.
- Integrations — exact, copy-paste configurations for opencode, Claude Code, GitHub Copilot in VS Code, Cursor, and Zed.
- Reference — supported databases, the model catalog, and the MCP tool catalog.
- Security & FAQ — encryption, sharing, retention, and the questions everyone asks.
A note on examples
API examples assume a key prefix of sk-coco- and the base URL https://api.cocobox.io/v1. Where a sample shows a key like sk-coco-1234abcd…, replace it with one you generated under API keys.
The Cocobox dashboard with a sidebar of database connections, an open SQL editor with AI suggestions, and a results table at the bottom.