Skip to Content

MCP Server

by
Odoo

$ 29.38

v 17.0 Third Party
Availability
Odoo Online
Odoo.sh
On Premise
Odoo Apps Dependencies Discuss (mail)
Lines of code 3934
Technical Name mcp_odoo_connector
LicenseLGPL-3
You bought this module and need support? Click here!
Availability
Odoo Online
Odoo.sh
On Premise
Odoo Apps Dependencies Discuss (mail)
Lines of code 3934
Technical Name mcp_odoo_connector
LicenseLGPL-3
MCP SERVER
Connect AI Assistants to Odoo ERP via Model Context Protocol

MCP Server is a powerful Odoo 17.0 module that securely connects AI assistants like Claude, Cursor, VS Code Copilot, Gemini, and ChatGPT to your Odoo ERP using the Model Context Protocol (MCP) standard. It provides REST API endpoints, XML-RPC proxy, and a built-in OWL Chatbot widget — all with fine-grained access control, full audit logging, rate limiting, response caching, and LLM-optimized output formatting.

Whether you need AI to read customer data, create sales orders, analyze inventory, or chat directly inside Odoo — MCP Server ensures every request is authenticated, authorized, logged, and delivered in a format AI assistants understand best.

  • ✅ REST API with 13 endpoints + XML-RPC proxy with 2 endpoints
  • ✅ Built-in OWL Chatbot widget (Ollama / OpenAI / Gemini)
  • ✅ Bearer Token (API Key) + Basic Auth support
  • ✅ Per-model CRUD access control with field-level restrictions
  • ✅ Rate limiting (configurable per minute per user)
  • ✅ IP whitelist for production security
  • ✅ Full audit logging with duration tracking
  • ✅ Response caching with configurable TTL
  • ✅ LLM-optimized output formatting
  • ✅ YOLO mode for development (bypasses permissions)
  • ✅ AI Function Calling — CRUD via chatbot
How to Use — Step by Step

Get up and running in 5 minutes

❶ Install the Module

  1. Copy the mcp_odoo_connector folder to your Odoo addons path.
  2. Restart Odoo server.
  3. Go to Apps → Update Apps List → Search "MCP Server" → Click Install.

❷ Enable MCP Server

  1. Go to Settings (top menu).
  2. Scroll to the MCP Server section.
  3. Toggle Enable MCP Server to ON.
  4. Click Save.

❸ Expose Models

By default, no models are exposed. You must explicitly add each model you want AI to access:

  1. Go to Settings → MCP Server → Model Access.
  2. Click New.
  3. Select a Model (e.g., Contact / res.partner).
  4. Toggle permissions: Read, Write, Create, Delete.
  5. (Optional) Set Allowed Fields as JSON: ["name", "email", "phone"]
  6. Click Save.
Model Technical Name Recommended Permissions
Contacts res.partner Read, Write, Create
Products product.template Read only
Sale Orders sale.order Read, Write, Create
Invoices account.move Read only
Tasks project.task Read, Write, Create
Employees hr.employee Read only

❹ Generate API Key

  1. Go to Settings → Users & Companies → Users.
  2. Select your user.
  3. Go to the API Keys tab.
  4. Click New API Key → Enter a description (e.g., "MCP Claude") → Scope: rpc.
  5. Copy the generated key — you will need it for AI client configuration.

❺ Install Python Dependencies (for MCP Bridge)

pip install "mcp[cli]" httpx

❻ Test the API

curl http://localhost:8069/mcp/api/v1/health

Expected response: {"status": "ok", "mcp_enabled": true, ...}

API Endpoints
➤ Public (No Auth Required)
Method Endpoint Description
GET /mcp/api/v1/health Health check with status, version, timestamp
GET /mcp/api/v1/system/info Odoo version, database, exposed model count
➤ Authenticated (Bearer Token or Basic Auth)
Method Endpoint Description
POST /mcp/api/v1/auth/validate Validate credentials, return user info
GET /mcp/api/v1/models List all exposed models with permissions
GET /mcp/api/v1/models/<model>/fields Get field metadata for a model
POST /mcp/api/v1/models/<model>/search Search records with domain filter
POST /mcp/api/v1/models/<model>/read Read specific records by IDs
POST /mcp/api/v1/models/<model>/browse Paginated browse with offset/limit
POST /mcp/api/v1/models/<model>/count Count records matching domain
POST /mcp/api/v1/models/<model>/create Create new record
POST /mcp/api/v1/models/<model>/write Update existing records
POST /mcp/api/v1/models/<model>/unlink Delete records
POST /mcp/api/v1/models/<model>/call Call arbitrary model method
➤ XML-RPC Endpoints
Endpoint Methods
/mcp/xmlrpc/2/common version(), authenticate()
/mcp/xmlrpc/2/object execute_kw()
➤ Usage Examples

Search Partners (curl):

curl -X POST http://localhost:8069/mcp/api/v1/models/res-partner/search \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"domain": [["is_company", "=", true]], "fields": ["name", "email"], "limit": 5}'

Note: Use dashes (res-partner) instead of dots (res.partner) in URL paths.

Create a Record (curl):

curl -X POST http://localhost:8069/mcp/api/v1/models/res-partner/create \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"values": {"name": "New Partner", "email": "new@example.com", "is_company": true}}'
AI Client Configuration

Connect your favorite AI assistant to Odoo using the included MCP Bridge script

Install dependencies first: pip install "mcp[cli]" httpx

💻 VS Code (Copilot / Claude)

Create .vscode/mcp.json in your workspace: 

{
  "servers": {
    "odoo": {
      "command": "python3",
      "args": ["/path/to/mcp_odoo_connector/mcp_bridge.py"],
      "env": {
        "ODOO_URL": "http://localhost:8069",
        "ODOO_API_KEY": "your-api-key-here",
        "ODOO_DB": "your-database-name"
      }
    }
  }
}
🤖 Claude Desktop

Config: ~/.config/claude/claude_desktop_config.json 

{
  "mcpServers": {
    "odoo": {
      "command": "python3",
      "args": ["/path/to/mcp_odoo_connector/mcp_bridge.py"],
      "env": {
        "ODOO_URL": "http://localhost:8069",
        "ODOO_API_KEY": "your-api-key-here",
        "ODOO_DB": "your-database-name"
      }
    }
  }
}
📡 Claude Code (CLI)

Run in your terminal:

claude mcp add-json odoo '{
  "command": "python3",
  "args": ["/path/to/mcp_odoo_connector/mcp_bridge.py"],
  "env": {
    "ODOO_URL": "http://localhost:8069",
    "ODOO_API_KEY": "your-api-key-here",
    "ODOO_DB": "your-database-name"
  }
}' -s user
🖱 Cursor IDE

Config: ~/.cursor/mcp.json

{
  "mcpServers": {
    "odoo": {
      "command": "python3",
      "args": ["/path/to/mcp_odoo_connector/mcp_bridge.py"],
      "env": {
        "ODOO_URL": "http://localhost:8069",
        "ODOO_API_KEY": "your-api-key-here",
        "ODOO_DB": "your-database-name"
      }
    }
  }
}

Replace /path/to/mcp_odoo_connector/mcp_bridge.py with the actual full path. ODOO_DB is required for multi-database servers.

🤖 Built-in AI Chatbot

Chat with AI directly inside Odoo — no external tools required

OpenAI

GPT-4o, GPT-4o-mini, and all OpenAI models. Enter your API key in settings and start chatting.

Google Gemini

Gemini Pro and Flash models. Paste your Google AI API key and connect instantly.

Ollama (Local)

Run LLMs locally — Llama 3, Mistral, Phi-3, and more. No API key needed. Full privacy.

🔧 AI Function Calling (Tool Use)

The chatbot can perform CRUD operations directly from the conversation:

  • "Create a new customer named John Doe" → AI creates the record in Odoo
  • "Update partner 14's phone to +1 555-0200" → AI updates the record
  • "Show me all unpaid invoices" → AI searches and returns results
  • "Delete test partner ID 99" → AI deletes the record
Security Best Practices
  • 🔒 Use HTTPS in Production — Always deploy behind a reverse proxy (Nginx/Caddy) with TLS certificates.
  • 🔑 Unique API Keys — Generate a separate API key for each AI integration. Revoke individually if compromised.
  • 🌐 Enable IP Whitelist — In production, restrict access to known IP addresses via MCP Server settings.
  • 🛡 Minimal Model Exposure — Only expose models your AI needs. Use field restrictions to hide sensitive columns.
  • ⛔ Disable YOLO Mode — YOLO mode bypasses all permission checks. Never enable it in production.
  • 📊 Review Audit Logs — Regularly check audit logs for suspicious activity. Auto-cleanup after 90 days (configurable).
Error Codes Reference
HTTP Error Code Meaning
401 AUTH_INVALID Missing or invalid credentials
403 BLOCKED_MODEL Model is security-blocked (ir.rule, etc.)
403 OPERATION_DENIED Operation not allowed on this model
404 MODEL_NOT_FOUND Model not exposed via MCP
429 RATE_LIMITED Rate limit exceeded — wait and retry
500 INTERNAL_ERROR Server error — check Odoo logs

MCP Server is an Odoo module that implements the Model Context Protocol standard, allowing AI assistants like Claude, Cursor, VS Code Copilot, and ChatGPT to securely read and write Odoo data through REST API, XML-RPC endpoints, and a built-in chatbot widget.

Any AI client that supports the MCP standard can connect, including Claude Desktop, Claude Code CLI, VS Code Copilot, and Cursor IDE. The built-in chatbot also supports OpenAI (GPT-4o), Google Gemini, and Ollama (local LLMs).

You can use either Bearer Token authentication (using Odoo's native API key system) or HTTP Basic Auth (login:password). API keys are recommended for production — generate them in Settings → Users → API Keys tab.

Yes — the module includes multiple security layers: Bearer token auth, IP whitelisting, per-user rate limiting, per-model CRUD permissions, field-level restrictions, blocked security-sensitive models, and complete audit logging. Just make sure YOLO mode is disabled and you use HTTPS.

YOLO mode is a development-only feature that bypasses MCP permission checks. It has two levels: "read_only" (skips permissions for read operations) and "full" (skips all permissions). Never enable it in production environments.

Use dashes instead of dots in URL paths. For example, use res-partner instead of res.partner. The API automatically converts dashes back to dots internally.

Yes! The built-in chatbot supports AI function calling (tool use) for all three providers — OpenAI, Gemini, and Ollama. The AI can create, read, update, and delete records directly from the chat interface.

Need Any Help?

Having trouble? We're happy to help you get started.

Developed & maintained by Mohamed Saied

📧 Contact us on Odoo Apps

Please log in to comment on this module

  • The author can leave a single reply to each comment.
  • This section is meant to ask simple questions or leave a rating. Every report of a problem experienced while using the module should be addressed to the author directly (refer to the following point).
  • If you want to start a discussion with the author or have a question related to your purchase, please use the support page.