mcp-salesforce-lite

Simple and lightweight Salesforce MCP server for connecting AI assistants to Salesforce data. Ideal for prototyping and small projects.

📦 Install from PyPI: pip install mcp-salesforce-lite

🔗 PyPI Package: https://pypi.org/project/mcp-salesforce-lite/

📚 GitHub Repository: https://github.com/luvl/mcp-salesforce-lite

Demo

See the MCP Salesforce Lite server in action with Claude Desktop:

The demo shows Claude Desktop using the MCP server to interact with Salesforce data - querying objects, retrieving records, and performing CRUD operations seamlessly.

Overview

This MCP (Model Context Protocol) server provides AI assistants like Claude with secure access to Salesforce data and operations. It implements the MCP standard to enable seamless integration between AI applications and Salesforce CRM.

Features

• 🔐 Secure Salesforce authentication via OAuth 2.0

• 📊 Access to Salesforce objects (Accounts, Contacts, Opportunities, etc.)

• 🔍 SOQL query execution

• 📝 CRUD operations on Salesforce records

• 🛡️ Built-in security and rate limiting

• 🚀 Easy setup and configuration

Quick Usage

Works with: Claude Desktop, any MCP-compatible AI assistant

Quick Start with Claude Desktop

Production Usage (Recommended)

The easiest way to use this MCP server is to install it directly from PyPI and configure it with Claude Desktop.

Step 1: Configure Claude Desktop

Add the following configuration to your Claude Desktop settings file:

Configuration File Location:

• macOS/Linux: ~/Library/Application Support/Claude/claude_desktop_config.json

• Windows: %APPDATA%\Claude\claude_desktop_config.json

Configuration:

Step 2: Set Up Salesforce Credentials

Replace the environment variables in the configuration:

• SALESFORCE_ACCESS_TOKEN: Your Salesforce access token

• SALESFORCE_INSTANCE_URL: Your Salesforce instance URL (e.g., https://yourcompany.my.salesforce.com)

Step 3: Restart Claude Desktop

After saving the configuration, restart Claude Desktop. You should see a hammer icon indicating that tools are available.

Step 4: Test the Integration

Try asking Claude:

• "List available Salesforce objects"

• "Describe the Account object"

• "Execute a SOQL query to get recent leads"

Prerequisites

• Python 3.10 or higher

• Salesforce Developer/Production org

• Connected App configured in Salesforce

Development Setup

If you want to modify or contribute to this MCP server, follow these development setup instructions.

Installation

Option 1: Using uv (Recommended for development)

Option 2: Using Poetry

Salesforce Development Setup

Create a .env file in the project root:

Usage

Development Mode

First, make sure you have your Salesforce credentials configured in your .env file.

Method 1: Direct Python Execution

Method 2: Using Poetry

Method 3: Using UV (Recommended)

Testing with MCP Inspector

If you have the MCP CLI installed, you can test your server:

How to Release the Server as a Pip Package

The server can be packaged and distributed via PyPI using the included pyproject.toml configuration.

Available Tools

The server provides the following tools that AI assistants can use:

Query Tools

• soql_query: Execute SOQL queries (schema must be defined to carefully ask for confirmation of UPDATE and DELETE operations)

• search_records: Search records across multiple objects with limit and pagination

• get_record: Retrieve a specific record by ID with limit and pagination

CRUD Operations

• create_record: Create new records (make sure to describe_object first, and find the reference fields of the objects)

• update_record: Update existing records

• delete_record: Delete records

Metadata Tools

• describe_object_definition: Get object metadata and field information with pagination

• list_avail_objects: List available Salesforce objects with limit and pagination

Development Claude Desktop Integration

If you're developing or running the server from source, you can use these alternative configurations:

💡 Tip: Example configuration files are provided in the examples/ directory:

• examples/claude_config_direct.json - Direct Python execution

• examples/claude_config_poetry.json - Poetry execution

• examples/claude_config_uv.json - UV execution (recommended)

Option 1: Direct Python Execution

Option 2: Poetry Execution

Option 3: UV Execution (Recommended for Development)

Project Structure

Release

Prerequisites

1. Register for PyPI Production: Go to https://pypi.org/account/register/

2. Enable 2FA: Set up two-factor authentication in your account settings

3. Create API Token: Go to https://pypi.org/manage/account/token/ and create a token

4. Update .pypirc: Replace pypi-YOUR_PRODUCTION_TOKEN_FROM_PYPI_ORG_HERE with your actual token

Publishing Process

1. Test on TestPyPI first:

2. Publish to Production PyPI:

Version Management

To publish a new version:

1. Update the version in pyproject.toml

2. Rebuild: uv build or poetry build

3. Upload: twine upload --repository pypi --config-file .pypirc dist/*