Traditional Odoo AI integrations often rely on a separate Python process communicating externally via XML-RPC or JSON-RPC. While functional, this approach introduces challenges like managing credentials in environment files, additional deployment overhead, and a lack of inherent visibility into AI actions within Odoo.
Addressing these limitations, the open-source addon muk_mcp transforms Odoo itself into a native MCP (Model-Controller-Presenter) server. This eliminates the need for extra processes, middleware, or RPC bridges. The /mcp endpoint resides directly within Odoo's registry and operates through the same ORM (Object-Relational Mapping) layer as all other Odoo functionalities.
After installing muk_mcp, users can generate an MCP key from their user preferences and paste its URL into compatible AI clients such as Claude Code, Claude Desktop, Cursor, Windsurf, Codex CLI, or OpenCode. The AI models gain access to 15 typed tools that cover the entire ORM lifecycle, including model discovery, schema introspection, search, read, create, update, delete, grouped aggregation, chatter interactions, and method execution.
Key Features and Benefits:
- Granular Permissions and Rate Limits: Each key supports model-level scopes with distinct read/write/create/delete permissions. Configurable rate limits per key are also available (defaulting to 60 requests/minute).
- Comprehensive Audit Log: Every AI call is logged, detailing the method, tool, model, duration, and status, ensuring full traceability of AI operations.
- UI-Based Custom Tool Definition: Custom tools can be added directly from the Odoo backend UI. Users define the tool's name, description, JSON Schema for inputs, and Python code executed within a sandboxed
safe_evalcontext. This eliminates the need for code deployment or server restarts, and connected clients are notified automatically when the tool list changes. - Session Management: Sessions maintain state according to the MCP specification and are automatically cleaned up after a configured timeout. Users also have the option to revoke their own sessions via user preferences.
- AI-Tagged Chatter: Any chatter message authored by an MCP tool is visibly tagged with a small MCP badge next to the author's name. This provides immediate clarity, distinguishing comments, status changes, and tracking entries generated by AI clients from those made by human users, enhancing transparency.
The introduction of muk_mcp provides a more efficient, secure, and transparent integration solution for leveraging AI Agents to directly drive Odoo operations, significantly simplifying the interaction between Odoo and AI models and advancing the intelligent transformation of enterprise management systems.