Salesforce DX MCP Server

Salesforce DX MCP Server is the official Model Context Protocol server from the Salesforce CLI organization for working with Salesforce orgs. The direct documentation link points to https://developer.salesforce.com/docs/atlas.en-us.sfdx_dev.meta/sfdx_dev/sfdx_dev_mcp.htm, while the repository https://github.com/salesforcecli/mcp contains the README, source code, license, and references to the @salesforce/mcp npm package. According to the README, the server is designed to let large language models read, manage, and operate Salesforce resources securely. For Skill Road, this entry fills the previously empty Lead Generation category because Salesforce data commonly includes contacts, accounts, leads, opportunities, and sales processes, even though the server itself comes from the developer toolchain.

CRM context for agents

The practical value appears when an agent can see not only code but also the structure of a Salesforce org. In sales, RevOps, and implementation projects, requirements often depend on objects, fields, flows, Apex code, permissions, and real CRM processes. Salesforce DX MCP Server can bring that context into an MCP client so an agent can inspect metadata, support standard DX tasks, prepare tests or scratch-org work, and reason about changes more accurately. This is especially useful where lead processes, campaign logic, CRM automation, or data quality in Salesforce meet code and configuration work.

Setup through Salesforce CLI

The server builds on the Salesforce DX developer toolchain. Before using it, teams need the Salesforce CLI installed and a Salesforce org already authorized. The MCP server is configured in a compatible client through the @salesforce/mcp npm package. The important point is that agents should not work with copied credentials. They use the existing Salesforce CLI authentication state and operate only within the authorized org and its permissions. For early testing, a sandbox or scratch org is far safer than a production sales environment.

Permissions, data, and governance

Salesforce data can contain personal information, leads, customer history, revenue expectations, contract context, or internal notes. Any MCP server that brings this data into an AI client must therefore be tightly controlled. Permissions follow the signed-in org, roles, profiles, permission sets, and Salesforce security models. Teams should decide in advance which objects and environments an agent may see, whether write actions are allowed, and how changes are reviewed. For lead generation and CRM analysis, a read-oriented first setup with a clear question is usually enough.

License, GitHub stars, and fit

The repository is Apache-2.0 licensed. The GitHub API reported exactly 467 stars for https://github.com/salesforcecli/mcp on 2026-09-07; that number is a snapshot of repository popularity, not a proof of quality or security. The server itself is open source, while Salesforce account access, org capabilities, platform functions, and the selected AI client follow their own provider terms. This public entry therefore does not state fixed prices. Lead Generation, Coding, and Automation fit because the server connects CRM objects and sales processes with development and automation tasks. Production use should start with a sandbox, minimal permissions, auditability, and human approval.

Salesforce CLI, authorized Salesforce org, Node.js/npm, and an MCP-compatible client.

Install Salesforce CLI, authorize an org, and configure @salesforce/mcp in the MCP client.

npx -y @salesforce/mcp

Existing Salesforce CLI authentication for the authorized org.

Permissions follow the Salesforce org, profiles, roles, permission sets, and enabled DX/MCP tools.

Tool calls run through the local DX/MCP toolchain and can return Salesforce metadata, CRM objects, and test results to the client.

Production CRM data, broad org rights, and unreviewed write actions can cause data exposure, faulty automation, or compliance issues.