Drizzle Docs

drizzle-docs-mcp

[](https://github.com/Michael-Obele/drizzle-docs/releases) [](https://opensource.org/licenses/MIT)

[](https://cursor.com/en-US/install-mcp?name=drizzle-docs-mcp&config=eyJjb21tYW5kIjoibnB4IC15IG1jcC1yZW1vdGUgaHR0cHM6Ly9kcml6emxlLm1hc3RyYS5jbG91ZC9hcGkvbWNwL2RyaXp6bGUtZG9jcy1tY3AvbWNwIn0%3D)

Mastra MCP server and tooling that provides real-time access to all Drizzle ORM documentation pages with fuzzy search, pre-caching, and flexible content retrieval.

Production Deployments

Choose the base host that fits your workflow — both expose the same toolset, but their runtime characteristics differ:

| Host | Base URL | Highlights | | ------------ | ---------------------------- | ---------------------------------------------------------------------------------------------------- | | Mastra Cloud | https://drizzle.mastra.cloud | Primary choice - Zero cold start, maximum responsiveness, and consistently reliable performance. |

• Append /api/mcp/drizzle-docs-mcp/sse for the SSE transport (best for editors that keep long-lived connections).
• Append /api/mcp/drizzle-docs-mcp/mcp for the HTTP transport (handy for CLIs and quick one-off calls).
• Mastra Cloud is the recommended primary deployment - it offers zero cold start and maximum responsiveness.

Endpoint reference & alternates

• Mastra Cloud SSE: https://drizzle.mastra.cloud/api/mcp/drizzle-docs-mcp/sse
• Mastra Cloud HTTP: https://drizzle.mastra.cloud/api/mcp/drizzle-docs-mcp/mcp

> [!NOTE] > This project follows our [Code of Conduct](CODEOFCONDUCT.md) and welcomes contributions! See our [Contributing Guidelines](CONTRIBUTING.md) for details.

This repository contains a Mastra-based MCP server that provides real-time access to Drizzle ORM documentation using a hybrid of pre-caching and live fuzzy search. Use it in your AI-powered code editor to get instant access to the latest Drizzle ORM patterns directly from the official docs.

Table of Contents

• [Production Deployments](#production-deployments)
• [Features](#-features)
• [Drizzle Ecosystem Integration](#drizzle-ecosystem-integration)
• [Observations & Minor UX Suggestions](#-observations--minor-ux-suggestions)
• [Editor Setup](#editor-setup)
• [CLI & Agent Configuration](#cli--agent-configuration)
• [Verification & Quick Tests](#verification--quick-tests)
• [Available Tools](#available-tools)
• [Example Usage](#example-usage)
• [Local Development](#local-development)
• [Developer Scripts](#developer-scripts)
• [MCP Architecture](#mcp-architecture)
• [Project Architecture](#project-architecture)
• [Contributing](#contributing)
• [License](#license)

🎉 Features

• ✅ Production deployment on Mastra Cloud
• ✅ Three main MCP tools for comprehensive Drizzle ORM support (see 'Available Tools')
• ✅ Smart Fuzzy Search: Powered by fuse.js, allowing for typos and partial matches.
• ✅ Pre-caching: Automatically fetches and indexes all 97 documentation pages at startup.
• ✅ Flexible Content Retrieval: Fetch full pages or specific sections in Markdown, JSON, or Plaintext.
• ✅ Support for all major AI code editors (Cursor, Windsurf, VS Code, Zed, Claude Code, Codex)
• ✅ HTTP and SSE transport protocols
• ✅ Real-time web scraping from orm.drizzle.team

Drizzle Ecosystem Integration

Drizzle ORM is more than just an ORM; it's a "Headless TypeScript ORM with a Head." This MCP server provides access to the entire ecosystem:

• Drizzle Kit: CLI for migrations, introspection, and Studio.
• Drizzle Relations (RQB): Best-in-class relational query builder.
• Validation Libraries: First-class support for zod, valibot, typebox, and arktype.
• Serverless/Edge Ready: Specific patterns for Turso, Neon, PlanetScale, and Cloudflare D1.

🔧 Observations & Suggestions

• The search_docs tool uses high-sensitivity fuzzy matching; if you get too many irrelevant results, try making your query more specific (e.g., "relational queries" instead of just "queries"). ✅
• Content is cached for performance, but the server will fetch fresh data if it detects significant upstream changes or if restarted. 💡
• The fetch_page tool allows for section-level extraction, which is highly recommended for keeping AI context windows clean.

Editor Setup

Mastra Cloud is the recommended primary deployment for all editors. It offers zero cold start and maximum responsiveness. VS Code users can open the Command Palette (Cmd/Ctrl+Shift+P) and run MCP: Add server to paste the URL.

Cursor

1. Open Cursor Settings (Cmd/Ctrl + ,).
2. Navigate to "MCP" / "Model Context Protocol".
3. Mastra Cloud is recommended. Append the SSE path as shown:

{
  "drizzle-docs": {
    "type": "sse",
    "url": "https://drizzle.mastra.cloud/api/mcp/drizzle-docs-mcp/sse"
  }
}

Windsurf

1. Edit ~/.codeium/windsurf/mcp_config.json.
2. Add the SSE transport as shown:

{
  "mcpServers": {
    "drizzle-docs": {
      "url": "https://drizzle.mastra.cloud/api/mcp/drizzle-docs-mcp/sse",
      "transport": "sse"
    }
  }
}

VS Code

Run MCP: Add Server (Ctrl/Cmd+Shift+P) and paste the URL:

• SSE: https://drizzle.mastra.cloud/api/mcp/drizzle-docs-mcp/sse

Available Tools

Once installed, your AI assistant will have access to these tools:

1. list_topics — Discover all 97 available Drizzle ORM documentation pages. Use this to understand the structure or find specific topic slugs.
2. fetch_page — Fetch and convert documentation pages to Markdown with optional filtering (slug, format, sections, maxLength).
3. search_docs — Search the documentation using intelligent fuzzy matching (query, limit).

Example Usage

Ask your AI assistant:

• "Show me how to setup a Postgres schema in Drizzle"
• "List all Drizzle docs topics"
• "Search for relational query examples in Drizzle"
• "How do I handle migrations with Drizzle Kit?"
• "Compare Drizzle select vs relational query builders"

Local Development

Quick start

1. Install dependencies:

bun install

1. Start development server (Mastra Studio):

bun run dev

1. Build for production:

bun run build

Developer Scripts

• npm run dev - Start Mastra Studio at localhost:4111.
• npm run build - Bundle the production-ready application.
• npm run mcp - Run the MCP server locally with TS support.
• npm run check - Verify TypeScript compilation.

Project Architecture

• Mastra Framework: Orchestrates the MCP server and tools.
• Smart Caching: Pre-fetches docs on startup to ensure instant search results.
• Turndown Service: Convers HTML content into clean, AI-optimized Markdown.
• Fuse.js: Handles typo-tolerant search across slugs and titles.
• Detailed Explanation: See [MCP_ARCHITECTURE.md](MCP_ARCHITECTURE.md).

License

This project is licensed under the [MIT License](LICENSE).

Contributing

We welcome contributions! Please read our [Contributing Guidelines](CONTRIBUTING.md) and [Code of Conduct](CODEOFCONDUCT.md) before getting started.

Contact

• Issues & Support: support@svelte-apps.me
• Contributions: contrib@svelte-apps.me
• Maintainer: Michael Amachree (michael@svelte-apps.me)

Source & license

This open-source MCP server is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: Michael-Obele
• Source: Michael-Obele/drizzle-docs
• License: MIT
• Homepage: https://mcp.svelte-apps.me/?mcp=drizzle-docs

Install and usage instructions live in the source repository linked above.