odoo

This MCP server integrates Odoo ERP with AI assistants, enabling both HTTP streaming access for remote AI agents and stdio access for local assistants. The HTTP-mode server exposes endpoints (via Docker Compose in the example) so that AI agents can send requests, receive streamed responses, and perform MCP operations like search, get, and create records against your Odoo instance. The stdio-mode server runs locally and exposes a module that accepts prompts fed through standard input, making it convenient for desktop AI tools. The included configuration demonstrates how to connect an MCP named odoo in Claude Desktop and how Flowhunt-like workflows can trigger Odoo operations through the MCP server.

Prerequisites:

• Docker and Docker Compose installed for HTTP mode
• Python 3.8+ installed for stdio mode
• Access to an Odoo instance (URL, database, user, and API key)

HTTP mode (Docker Compose):

1. Clone the repository and navigate to it

git clone https://github.com/vzeman/odoo-mcp-server.git
cd odoo-mcp-server

2. Copy the example docker-compose file and configure credentials in docker-compose.yml as needed

cp docker-compose.example.yml docker-compose.yml

3. Start the HTTP MCP server

docker-compose up -d

4. Health check

curl http://localhost:8000/health

stdio mode (Python):

1. Install from PyPI (recommended) or clone the repo and install in development mode

pip install odoo-mcp-server  # or
# or from source:
# git clone https://github.com/vzeman/odoo-mcp-server.git
# cd odoo-mcp-server
# pip install -e .

2. Set environment variables for Odoo access as described in the README

export ODOO_URL="https://your-instance.odoo.com"
export ODOO_DB="your-database"
export ODOO_USERNAME="your-email@example.com"
export ODOO_API_KEY="your-api-key"

3. Run the HTTP server (optional, if you want the HTTP interface in stdio install path)

python -m mcp_server_odoo.http_server

4. Run the stdio server for local assistants

python -m mcp_server_odoo

Environment variables required by the Odoo MCP server are listed in the README and the Claude Desktop configuration example. When using ngrok for remote access, ensure your public endpoint is secured and that the MCP server is reachable from Flowhunt or other AI agents. If you encounter connection issues, verify that ODOO_URL, ODOO_DB, and credentials are correct, and that the MCP server has network access to the Odoo instance. For Docker-based HTTP mode, ensure Docker Compose is using the same ports referenced in health checks (default 8000). The stdio mode is suitable for local prototyping with Claude Desktop or other local assistants.