dmtoleadDocs
TRESStart free
DocsOperate with AI clientsRemote MCP

Connect an AI operator with MCP

Give Codex, Claude Code or another remote MCP client tenant-scoped tools to inspect, configure, privately test and explicitly control a DMtoLead workspace.

DMtoLead product view
Current UI
Remote AI operator access
Connect an AI operator with MCP

Give Codex, Claude Code or another remote MCP client tenant-scoped tools to inspect, configure, privately test and explicitly control a DMtoLead workspace.

Remote MCPhttps://api.dmtolead.com/mcp
workspace.statusLive workspace and next gateRead
setup.planOrdered safe setup pathRead
assistants.test_privateOne invisible replyPrivate
assistants.activateExplicit customer-facing startLocked
Permission boundaries
01
dmtolead.read

Inspect workspace facts

02
dmtolead.setup

Configure while paused

03
dmtolead.activate

Start or pause live replies

Code-rendered product view based on the current DMtoLead interface. Example data only.

Setup does not mean go live
MCP setup tools keep customer-facing replies paused. Activation requires its own OAuth permission, a current private test you explicitly accepted, and the exact confirmation START AUTOMATIC REPLIES.

What the MCP connection exposes

01

Twenty scoped tools

Read workspace status, activity, customers, business data, usage and handoffs; then create a brain, choose an outcome, teach essentials, bind a channel and run the reviewed activation loop.

02

Three permission boundaries

Read access, paused setup and customer-facing runtime control are separate OAuth scopes. The consent screen shows each boundary before access is granted.

03

Credentials stay in DMtoLead

The client receives safe tool results—not Meta, Telegram, Shopify or other provider tokens. Every call resolves to the signed-in workspace and records an audit reference.

Connect a remote MCP client

  1. 1

    Open Settings → MCP

    Choose Codex, Claude Code or OpenCode. The page shows the stage-aware server URL, exact setup commands and the three requested permissions.

  2. 2

    Add the remote server

    For Codex, run: codex mcp add dmtolead --url https://api.dmtolead.com/mcp. Other clients use the same remote HTTP endpoint.

  3. 3

    Sign in and approve the scopes

    Run: codex mcp login dmtolead --scopes dmtolead.read,dmtolead.setup,dmtolead.activate. DMtoLead opens its own login and consent screen using OAuth 2.1 authorization code flow with S256 PKCE.

  4. 4

    Start with a read-only request

    Ask for workspace status and the next safe setup step. Review the plan before authorizing configuration or customer-facing runtime changes.

A safe first prompt

This calls only the tenant-scoped read path and returns the current setup gate without changing the workspace.

Show me my DMtoLead workspace status and the next safe setup step.

Security and operating notes

  • The production resource URL is https://api.dmtolead.com/mcp; local dashboard builds show the development endpoint and server name dmtolead_dev.
  • Private test replies are customer-invisible and intentionally one-time. Acceptance is bound to that exact tested configuration.
  • assistants.pause is immediate and preserves configuration, channel authorization and accepted test evidence.
  • OAuth access is workspace-scoped. Managed-account switching is not passed through the private MCP boundary.
Related guides
AI Assistants
Connect a customer account, choose the job DMtoLead should handle, train it with approved knowledge and activate only after testing.
Your dmtolead Assistant
Turn a private conversation you own into an operator channel for live account facts, customer alerts, calendar actions and safe prompt improvements.
Inbox, testing & handoff
Test replies, review conversations and take over cleanly.

Need help? Message us in the website chat or email support@dmtolead.com. dmtolead.com

Previous
Outbound webhooks
Next
Pricing & credits