Which integrations are available for this server?

Provides tools to interact with Odoo resources programmatically via ODXProxy, enabling management of Odoo instances and data.

How do I use odxproxy-mcpserver?

1. Click on "Install Server". 2. Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state. 3. In the chat, type @ followed by the MCP server name and your instructions, e.g., "@odxproxy-mcpserver list all products in Odoo" 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.

ODXProxy MCP Server

Official MCP Server for ODXProxy by TERRAKERNEL. PTE. LTD.

Overview

This package provides a Model Context Protocol (MCP) server that connects ODXProxy with MCP-compatible clients/tools. It exposes capabilities to interact with ODX/Odoo resources programmatically.

Status

Early version (0.1.x). Expect breaking changes as the APIs evolve.

Usage

See the entry points in src/index.ts and src/server.ts. Typical usage is to import the built output from dist after building.

import {OdxInstanceInfo, OdxMCPServer,OdxProxyClientInfo} from "odxproxy-mcpserver";
import { StdioServerTransport } from "@modelcontextprotocol/sdk/server/stdio.js";


async function main() {
    const env = process.env;
    let clientInfo: OdxProxyClientInfo = {
        instance: {
            url: "ODOO_INSTANCE_URL",
            user_id: 2,
            db: "DB_NAME",
            api_key: "YOUR_ODOO_API_KEY"
        },
        odx_api_key: "YOUR_ODXPROXY_API_KEY",
        gateway_url: "https://gateway.odxproxy.io"
    };
    let server = new OdxMCPServer(clientInfo);
    await server.initBaseResource();

    // If DRY_RUN is set, initialize and exit to allow smoke testing without stdio hang.
    if (env.DRY_RUN === "1") {
        return;
    }
    let transport = new StdioServerTransport();
    await server.connect(transport);
}

main().catch((error) => {
    console.error(error);
});

You can also build the package locally and run from source.

Scripts

build: Build ESM/CJS and type declarations with tsup ~~- test: Run Jest in CI mode with JUnit reporter~~ ~~- test:watch: Watch mode for Jest~~ ~~- prepublishOnly: Build and test before publishing~~

Run with:

npm run build

After you successfully build the package, embed the built output in your project to any supported LLM client like Claude Desktop or your own MCP Client. Visit https://modelcontextprotocol.io for more information about connecting to local MCP Servers