Odoo MCP Server

by Cybrosys Techno Solutions

v 19.0 Third Party 98

Versions

18.0 19.0

You bought this module and need support? Click here!

Versions

18.0 19.0

Description
Manifest

Supports: Community Enterprise

Availability:

On Premise

Odoo.sh

Odoo Online

NEW · v19.0.2.0.0 Per-user tool execution, a model access allow-list, hardened webhooks & constant-time auth

Productivity · AI Gateway

Odoo MCPServer

Connect AI assistants (Claude, Cursor, Codex) and chat bots (Telegram, WhatsApp, Discord, Web) to your Odoo data through a single, secure Model Context Protocol gateway.

Email Us

+91 9074270811

Instant Support | https://support.cybrosys.com/instant-support

Read the Documentation

Step-by-step setup guides for every AI client, provider and chat-bot platform ship with the module in DOCUMENTATION.md.

Visit Cybrosys

Get Expert Support

Need help connecting a client or provider? Our team is ready — reach out any time by email or WhatsApp.

Contact Support

Explore Features In Detail

Pick a category to see its detailed features.

MCP Server

A full Streamable-HTTP JSON-RPC endpoint at /mcp_gateway that any MCP-capable client can connect to with an Odoo API key.

1	Initialize Handshake Full MCP handshake with client identification and reconnect support.

2	tools/list & tools/call Discover every active tool and execute it with typed arguments.

3	Resources (odoo://) Expose Odoo models as readable resources via the odoo:// URI scheme.

4	Protocol Sessions Track each connected client with its own session and reconnect logic.

5	API-Key Authentication Bearer token or ?api_key= — validated against Odoo's own API keys.

6	Health Endpoint A public /mcp_gateway/health check to confirm the gateway is live.

AI Providers

Configure one or more providers; the gateway picks the highest-priority active one and falls back automatically if it fails.

1	OpenAI GPT-4o and friends for chat completion and embeddings.

2	Anthropic (Claude) Opus, Sonnet and Haiku with large context windows.

3	Google Gemini Gemini 2.5 Flash / Pro with million-token context.

4	Ollama (Local) Run Llama, Mistral, Gemma and more fully on-premise.

5	Custom OpenAI-Compatible Groq, Together, LM Studio, vLLM — any /v1 endpoint.

6	Fetch Models & Fallback Auto-import model lists; priority ordering with automatic fallback.

Tools & Framework

Turn any Odoo operation into an AI-callable tool — with auto-generated JSON schemas.

1	@ai_tool Decorator Register any model method as a tool; schema built from type hints.

2	Built-in ORM Tools search, create, update, delete and unlink records generically.

3	ask_ai Send a free-form prompt to the configured provider and get a reply.

4	analyze_records Fetch records and ask the AI to analyse them in one call.

5	Per-Tool Provider / Model Point each tool at a default provider and model of your choice.

6	Test Tool Wizard Fire live parameters at any tool and inspect the real output.

Chat Bots

Let users talk to Odoo in natural language from the platforms they already use.

1	Telegram Paste your BotFather token and Connect — webhook auto-registered.

2	WhatsApp Cloud API Connect a Meta business number with the verify handshake.

3	Discord Interactions endpoint with verified Ed25519 request signatures.

4	Web Widget A plain JSON endpoint you can call from any site or app.

5	Conversation Memory Per-user chat history so the bot remembers context.

6	LLM Intent Routing The bot understands the message and calls the right tool for it.

Security

Layered controls keep the AI inside the boundaries you set.

1	Per-User Execution Tools run as the API-key's user, honouring ACLs and record rules.

2	Model Access Allow-list Choose exactly which models & operations tools may use — destructive off by default.

3	User Consent Gate Flag sensitive tools to require approval before every run.

4	Webhook Authentication Shared secret plus verified Discord Ed25519 signatures.

5	Protected Credentials Provider keys, bot tokens & secrets restricted to the System group.

6	Rate Limiting & Caps Per-IP rate limits and request body-size caps on every endpoint.

Dashboard & Logs

Stay in control with a live control panel and a full audit trail.

1	AI Hub Dashboard Server status, providers, tools, sessions and consents at a glance.

2	Tool Logs Every call logged with user, source, model, duration and result.

3	Protocol Sessions See which clients are connected and when they were last active.

4	AI-Ready Messaging mail.message is extended with roles and structured tool-call data.

Before Installation

This module uses a few external Python packages. Install them first, then install the module.

Required packages: pydantic, mcp, jinja2 and requests. Install these before installing the module using the commands below.

Install the Python packages

bash

pip3 install pydantic
pip3 install mcp
pip3 install jinja2
pip3 install requests

Key Highlights

Everything the gateway brings to Odoo 19 — at a glance.

Full MCP Server

Streamable-HTTP JSON-RPC endpoint for Claude, Cursor, Codex and more.

5 AI Providers

OpenAI, Anthropic, Gemini, Ollama and any OpenAI-compatible endpoint.

Automatic Fallback

Priority ordering with fallback to the next provider if one fails.

Tool Framework

@ai_tool decorator plus generic ORM, ask_ai and analyze_records tools.

4 Chat Channels

Telegram, WhatsApp, Discord and a generic Web widget with memory.

Per-User Security

Tools honour each user's own Odoo access rights and record rules.

Access Allow-list

Admin-controlled models & operations; destructive actions off by default.

User Consent Gate

Require an approver to authorise sensitive tools before they run.

Full Tool Logging

Audit every call — user, source, model, duration, status and result.

AI Hub Dashboard

Live counts, server status and quick actions on a single screen.

Model Resources

Expose Odoo models as readable resources via the odoo:// URI scheme.

Key Generation Wizard

Generate a client-ready API key and config snippet per platform.

App Screenshots

Dashboard & Tools Providers & Models Resources Connect AI Clients Consent & Logs Chat Bots

AI Hub Dashboard — Central Control Panel

AI Tools — Define Callable Actions

Browse & Test MCP Tools

Tool Logs — Full Audit Trail

AI Providers

Fetch Available Models

MCP Exposed Resources

Generate an MCP API Key

Connect Claude Code CLI

Connect claude.ai (Cloud)

Connect Antigravity

AI Tool Consent Granting

Set Up Bot Channels

Detailed Features

A complete AI gateway — from the MCP server to providers, tools, bots and security.

01 MCP JSON-RPC Server Streamable-HTTP endpoint at /mcp_gateway.	02 Multi-Client Support Claude Desktop, Claude Code, claude.ai, Cursor, Codex.	03 5 AI Providers OpenAI, Anthropic, Gemini, Ollama, custom.
04 Provider Fallback Priority ordering with automatic fallback.	05 Fetch Models Auto-import model lists from the provider API.	06 @ai_tool Decorator Register model methods as tools automatically.
07 Built-in ORM Tools search, create, update, delete, unlink.	08 ask_ai & analyze_records Prompt the AI or have it analyse Odoo data.	09 Model Resources Expose models via the odoo:// URI scheme.
10 Telegram Bot Auto webhook registration from a bot token.	11 WhatsApp & Discord Cloud API verify flow; Discord Ed25519 signatures.	12 Web Chat Widget A plain JSON endpoint for any site or app.
13 Conversation Memory Per-user chat history for contextual replies.	14 Per-User Execution Tools honour each user's ACLs and record rules.	15 Access Allow-list Per-model, per-operation control; destructive off by default.
16 Consent Gate Approval required for sensitive tools.	17 Tool Logging Full audit trail for every invocation.	18 AI Hub Dashboard Live stats and quick actions on one screen.

Frequently Asked Questions

What is MCP and which clients can connect?

MCP (Model Context Protocol) is an open standard for connecting AI assistants to tools and data. Any MCP-capable client works — Claude Desktop, Claude Code CLI, claude.ai, Cursor and OpenAI Codex are all supported out of the box.

Which AI providers are supported?

OpenAI, Anthropic (Claude), Google Gemini, Ollama (local), and any OpenAI-compatible endpoint such as Groq, Together AI, LM Studio or vLLM. You can configure several and the gateway falls back automatically.

Can the AI change or delete my data?

Only if you allow it. Tools run with the connecting user's own access rights, and a Tool Access allow-list controls exactly which models and operations are permitted. Reads on common models are enabled by default; create, update and delete are off until an administrator turns them on.

Which chat platforms can I connect?

Telegram, WhatsApp Cloud API, Discord and a generic Web widget. Each bot understands natural language, calls the right Odoo tool and replies — with per-user conversation memory.

Which Python packages are required?

pydantic, mcp, jinja2 and requests. Install them with pip before installing the module (see the Before Installation section above).

Does it work on Odoo Online?

It runs on On-Premise and Odoo.sh deployments (the external Python packages are added via requirements.txt on Odoo.sh). Odoo Online (SaaS) does not allow custom modules with external dependencies, so it is not supported there.

Latest Releases

Latest Release 19.0.2.0.0

IMP

27th Jul, 2026

Security refactor: per-user tool execution, model access allow-list, hardened webhooks and constant-time authentication.

Release 19.0.1.0.0

Add

Initial Release

Initial release: MCP server, AI providers, tool framework, chat bots, consent, sessions and dashboard.

Our Services

Odoo Customization

Odoo Implementation

Odoo Support

Odoo Migration

Odoo Integration

Odoo Consultancy

Odoo Licensing

Hire Odoo Developer

Related Modules

Odoo 19 Full Accounting Kit

Open HRMS Core Odoo 19

Inventory Barcode Scanning

Availability	Odoo Online Odoo.sh On Premise
Odoo Apps Dependencies	Discuss (mail)
Lines of code	5823
Technical Name	`odoo_mcp_manager`
License	LGPL-3
Website	https://www.cybrosys.com

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, please use the developer contact information. They can usually be found in the description.