Custom-built MCP Server for Salesforce

Описание и применение

Сервер MCP, развернутый локально или в Docker, позволяет AI‑агентам выполнять CRUD, запросы и другие операции в Salesforce через стандартизированные команды, используя OAuth PKCE или клиентские учетные данные.

Задача

Enable AI agents to query, search, and modify Salesforce data via a self-hosted MCP server.

Результат

The server enables AI agents to perform Salesforce CRUD, query, bulk, and custom API operations via MCP commands, with optional authentication flows and retry logic.

Рабочий процесс

• Set up a Salesforce org with an External Client App and obtain Consumer Key
• Choose a deployment option: Option A (uvx), Option B (Python from source), or Option C (Docker)
• Configure environment variables SF_LOGIN_URL and SF_CLIENT_ID (and optionally SF_CLIENT_SECRET, SF_AUTH_FLOW, SF_PKCE_TOKEN_CACHE)
• Run one-time interactive login with sf-mcp-login (or use client credentials flow)
• Start the server with sf-mcp-server (or docker run)
• Configure an MCP client (e.g., Claude Desktop) to use the server command and env vars
• Use MCP tools like sf_query, sf_create_record, etc. to interact with Salesforce

Доказательства

A custom-built, self-hosted MCP server that lets AI agents (Claude Desktop, Claude Code, the MCP Inspector, or any other MCP client) connect to Salesforce — to query, search, and modify data in an org.

Источник

Three ways to get a running server — pick whichever fits: Option A — uvx, Option B — Python from source, Option C — Docker

Источник

The server also runs as a container behind a network-reachable Streamable HTTP endpoint, for when you want an agent that isn't on the same machine to reach it.

Источник