What MCP tools does Serpent expose?

plan_deploy, create_pull_request, resolve_metadata_conflict, and trigger_pipeline, with more tools shipping as the protocol matures.

Do I need to install anything to use MCP with Serpent?

No install in your Salesforce org. Point your AI client at Serpent's MCP server and you're connected.

Does MCP work across multiple orgs?

Yes. Serpent's MCP server spans your full org landscape, so you can scope, review, and deploy across sandboxes and production from one AI client.

Is MCP available on the free plan?

Yes. MCP access is included on Serpent's free Essentials tier, no credit card required.

The only native MCP server in Salesforce DevOps

Salesforce MCP Server

Deploy, open pull requests, resolve conflicts, and trigger pipelines straight from Claude, Cursor, Copilot, or Codex.

Start free Book a demo

Connection details

Connect an MCP client to Serpent

Point any MCP client at the endpoint below and sign in with OAuth.

Server details

Endpoint: https://mcp.serpent.cloud/mcp
Transport: Streamable HTTP
Authentication: OAuth 2.1 with PKCE (S256) and dynamic client registration, so your client opens a browser to consent and no API key ever sits in the config.
Scope: serpent:mcp

Client configuration

{
  "mcpServers": {
    "serpent": {
      "type": "http",
      "url": "https://mcp.serpent.cloud/mcp"
    }
  }
}

That is the Claude Code and Cursor shape. Every client speaks the same endpoint and transport, but the config key names differ, so use your own client's page: Claude, Cursor, Codex, Cline, Windsurf or GitHub Copilot.

MCP tools

What you can do from your AI client

Four MCP tools give your AI assistant real Salesforce DevOps actions, not just chat.

Plan a deployment

The plan_deploy tool scopes a deployment and maps the exact components before anything moves.

Open a pull request

The create_pull_request tool opens a reviewed PR straight from your AI client, no browser needed.

Resolve conflicts

The resolve_metadata_conflict tool resolves metadata conflicts, escalating only what needs your judgment.

Trigger a pipeline

The trigger_pipeline tool kicks off a pipeline run and reports status back in your AI client.

From prompt to production

How it works

Three steps from a plain-English request to a deployed change.

Connect your AI client

Point Claude, Cursor, Copilot, or Codex at Serpent's MCP server and sign in with OAuth. No new interface, no new credentials.

Ask in plain English

Tell your assistant what to ship, review, or fix. Serpent's MCP tools translate intent into scoped actions.

Review, then ship

Every MCP-triggered action still runs through Serpent's preflight checks, AI review, and mandatory human approval before it executes.

Common questions

Salesforce MCP server, answered

How the native MCP server works, which AI clients it supports, and what you can run from them.

What is a Salesforce MCP server?

MCP (Model Context Protocol) lets AI assistants call real tools instead of just chatting. Serpent's MCP server exposes Salesforce DevOps actions directly to your AI client.

What is the MCP connection URL?

It is https://mcp.serpent.cloud/mcp, over streamable HTTP with OAuth 2.1 sign-in on scope serpent:mcp. Each AI client's integration page carries its own setup snippet.

Which AI clients does Serpent's MCP server support?

Claude, Cursor, Cline, Codex, Windsurf, and GitHub Copilot today, with more clients added as MCP adoption grows.

Is Serpent really the only native MCP server for Salesforce DevOps?

Yes. Serpent is the only Salesforce DevOps platform shipping a native MCP server today.

Can the AI deploy to production without me?

No. Every MCP-triggered deployment still requires your human approval before anything ships.

Start free. No credit card, no install, no commitment.

Connect your AI client and ship your first guarded deployment today.