DataToRAG MCP
OfficialProvides tools for interacting with Atlassian products, including Jira and Confluence, enabling issue and page management, searching, and commenting.
Allows reading and editing pages, searching by CQL, managing comments, and attachments.
Provides tools for searching threads, reading and composing emails, managing labels, and drafting replies.
Allows searching by JQL, creating and transitioning issues, managing comments and attachments.
Click on "Install Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@DataToRAG MCPsearch my Gmail for invoices from March"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
DataToRAG MCP
Connect your Google Workspace, Jira, and Confluence to Claude — through one MCP endpoint.
DataToRAG is an open-source Model Context Protocol gateway that turns your everyday work tools into something your AI assistant can actually use. Gmail, Calendar, Drive, Docs, Sheets, Slides, Tasks, Contacts, Jira, and Confluence — 70 tools across one endpoint, with multi-account support and token-optimized responses.
Website → datatorag.com
Docs → datatorag.com/docs
Dashboard → datatorag.com/dashboard
Why DataToRAG
Works with any MCP client. Claude Desktop, Cursor, Windsurf, custom agents — anything that speaks the Model Context Protocol.
Google Workspace, natively. 48 tools across Gmail, Calendar, Drive, Docs, Sheets, Slides, Contacts, and Tasks. Search threads, draft replies, update cells, batch-create slides, find files, manage tasks — from a single prompt.
Atlassian, too. 22 tools for Jira and Confluence: search by JQL or CQL, create and transition issues, read and edit pages, manage comments and attachments.
Multi-account out of the box. Connect personal, shared, and team Google accounts under one endpoint. Your assistant can target a specific account or search across all of them.
Token-optimized. Naive API wrappers dump everything into your context window. DataToRAG tunes tool responses for token efficiency — the same Gmail thread read costs a fraction of the tokens, which means longer conversations and smarter agents.
OAuth per user. Each user signs in with Google and Atlassian themselves. No shared service accounts, no tokens in code.
Self-hostable. MIT-licensed. Run it on your own infrastructure, or use the hosted version at datatorag.com.
Related MCP server: MCP-GET
Quick Start (hosted)
Add this to your MCP client config:
{
"mcpServers": {
"datatorag": {
"url": "https://datatorag.com/mcp"
}
}
}Then sign in at datatorag.com/dashboard and connect your Google or Atlassian accounts. That's it — your AI assistant now has access to every tool listed in the docs.
Available Connectors
Connector | Tools | Services |
48 | Gmail · Calendar · Drive · Docs · Sheets · Slides · Contacts · Tasks | |
22 | Jira · Confluence |
Each tool is documented with its operations, required OAuth scopes, and example prompts at datatorag.com/docs.
Architecture
┌──────────────────┐
│ MCP Client │
│ Claude Desktop, │
│ Cursor, etc. │
└────────┬─────────┘
│
POST /mcp │ Bearer <OAuth token>
│
▼
┌──────────────────┐
│ 1. Auth │── OAuth2 token
│ │ validation
└────────┬─────────┘ 401 if invalid
│
▼
┌──────────────────┐
│ 2. Session │── Streamable HTTP
│ │ Create or reuse
└────────┬─────────┘
│
▼
┌──────────────────┐ ┌────────────┐
│ 3. tools/list │───>│ PostgreSQL │
│ tools/call │ │ │
└────────┬─────────┘ │ users │
│ │ tools │
▼ │ accounts │
┌──────────────────┐ │ │
│ 4. Route by │<──>│ │
│ namespace │ └────────────┘
└────────┬─────────┘
│
┌────────────┼─────────────┐
▼ ▼ ▼
┌────────────┐ ┌────────────┐ ┌──────────┐
│ gws-mcp │ │ atlassian- │ │ plugin-n │
│ (Node.js) │ │ mcp │ │ ... │
│ :40000/mcp │ │ :40001/mcp │ │ │
└────────────┘ └────────────┘ └──────────┘The gateway is a thin routing layer: it authenticates the client, forwards per-user OAuth tokens to the plugin that owns the tool namespace, and streams the response back. Plugins run as separate Node.js processes — they can be added, removed, or replaced without changing the gateway.
Self-Hosting
Prerequisites
Docker
Google OAuth client for login (and a separate one for Google Workspace scopes)
Atlassian OAuth client (optional, for Jira + Confluence)
Local Development
# Start postgres, run migrations, seed a test user, and boot the gateway
docker compose -f docker/docker-compose.dev.yml up -dService | URL |
Gateway | |
Dashboard | |
Docs | |
Health | |
OAuth metadata | http://localhost:8285/.well-known/oauth-authorization-server |
Connect a local MCP client:
{
"mcpServers": {
"datatorag": {
"url": "http://localhost:8285/mcp"
}
}
}Development Without Docker
# Start only postgres
docker compose -f docker/docker-compose.dev.yml up postgres -d
pnpm install
# Push schema + seed
DATABASE_URL=postgresql://datatoragmcp:localdev@localhost:54320/datatoragmcp \
pnpm --filter @datatorag-mcp/db db:push && \
pnpm --filter @datatorag-mcp/db db:seed
# Start gateway
DATABASE_URL=postgresql://datatoragmcp:localdev@localhost:54320/datatoragmcp \
pnpm --filter @datatorag-mcp/gateway devEnvironment Variables
Variable | Default | Required | Description |
| — | Yes | PostgreSQL connection string |
|
| No | Gateway server port |
|
| No | Public URL for OAuth redirects |
| — | For login | Google OAuth client ID (dashboard sign-in, minimal scopes) |
| — | For login | Google OAuth client secret |
| — | For GWS tools | Separate OAuth client for Google Workspace scopes |
| — | For GWS tools | — |
| — | For Atlassian tools | Atlassian OAuth client ID |
| — | For Atlassian tools | — |
| — | No | PostHog project key for server-side analytics (tool calls, signups, OAuth) |
| — | No | Same value as |
Production Deployment
Production runs on Docker Compose with a build-arg pipeline for NEXT_PUBLIC_* values. See docker/docker-compose.prod.yml and apps/gateway/Dockerfile for the reference setup.
Project Structure
apps/
gateway/ MCP proxy server (Express + MCP SDK) + Next.js dashboard
src/
app/ Dashboard, docs, landing page (Next.js App Router)
gateway/ MCP routing, auth, plugin manager, tool tracking
lib/ Shared utilities (docs parser, analytics, hooks)
content/
blog/ Markdown blog posts
docs/ Markdown connector docs
server.ts Custom Express server entry point
packages/
config/ Environment variable parsing (zod schema)
db/ Drizzle ORM schema + database client
types/ Shared TypeScript types
auth/ OAuth helpers
docker-manager/ Docker lifecycle helpers for plugin containers
docker/
docker-compose.dev.yml Local development stack
docker-compose.prod.yml Production (Lightsail)Contributing
Issues and pull requests welcome. For substantial changes, open an issue first so we can align on direction.
License
MIT
This server cannot be installed
Maintenance
Resources
Unclaimed servers have limited discoverability.
Looking for Admin?
If you are the server author, to access and configure the admin panel.
Latest Blog Posts
- Your AI Chatbot Just Exposed Your CEO's Salary to an InternBy Om-Shree-0709 on .Agent IdentityMCP SecurityOAuth Delegation
- Why MCP Servers Need Execution Sandboxing (And Why Your Current Stack Isn't Enough)By Om-Shree-0709 on .Agentic AiPrompt InjectionWebAssembly
MCP directory API
We provide all the information about MCP servers via our MCP API.
curl -X GET 'https://glama.ai/api/mcp/v1/servers/datatorag/mcp-gateway'
If you have feedback or need assistance with the MCP directory API, please join our Discord server