home / mcp / salesforce documentation mcp server

Salesforce Documentation MCP Server

Provides offline access to Salesforce docs by crawling, indexing, and searching through local content for quick answers.

Installation

Add the following to your MCP client configuration file.

Configuration

View docs

{
  "mcpServers": {
    "ttrevisan-ilmn-unified-sf-docs-mcp": {
      "command": "npx",
      "args": [
        "-y",
        "unified-sf-docs-mcp"
      ]
    }
  }
}

You can run a focused MCP server that lets your AI assistants scrape, index, and search Salesforce documentation, handling complex Shadow DOMs and legacy iframe structures for fast local queries.

How to use

Connect this MCP server to your AI assistant to enable instant access to Salesforce documentation. Start the MCP server from your local environment, then configure your assistant to point at the running process. Use the three core capabilities to gather, index, and query content: scrape a single Salesforce doc page, crawl a central guide to build a local doc bundle, and search the local index for precise answers.

How to install

Prerequisites: you need Node.js and npm installed on your machine.

Step 1: Clone the repository and navigate into the project directory.

git clone https://github.com/ttrevisan-ilmn/unified-sf-docs-mcp.git
cd unified-sf-docs-mcp

Step 2: Install dependencies.

npm install

Step 3: Build the project.

npm run build

Step 4: Run or integrate the MCP server with your AI assistant. The simplest interactive usage is to start the MCP via the package command shown for MCP clients.

npx -y unified-sf-docs-mcp

Additional notes

If you prefer local control without npx, you can integrate into your workflow by running the built output from the dist directory. Maintain the required command and arguments exactly as shown in your integration setup.

Available tools

scrape_single_page

Load a Salesforce documentation URL with a headless browser, pierce all shadow DOMs, wait for dynamic content, and return clean Markdown for indexing.

mass_extract_guide

Provide a central guide URL to automatically extract the parent page, discover hierarchical child links, scrape them concurrently, chunk their content, and store to a local SQLite database for offline querying.

search_local_docs

Query the local SQLite index with a natural language prompt to retrieve the best matching documentation chunks instantly.