Workflows Added to LeadConnector MCP Server in Private Beta

TL;DR

HighLevel added workflow support in private beta to the LeadConnector MCP server for AI tools like Claude, ChatGPT, and Cursor. You only need to act if you run AI assistants and have the Workflows API enabled on your sub-accounts.

Who this affects

HighLevel has not confirmed which specific plans or users are affected. Access requires sub-accounts to have the Workflows API enabled. Agencies not connecting external AI tools to their accounts can ignore this update.

What changed

• Added workflow support in private beta to the LeadConnector Model Context Protocol (MCP) server for AI tools including Claude, ChatGPT, and Cursor.
• Enabled searching workflows by name, status, folder, or date through MCP.
• Enabled reading full workflow definitions, including triggers, steps, branches, and settings.
• Enabled building workflows from text descriptions with error checking, saving them directly as drafts.
• Enabled updating, publishing, reverting to draft, and deleting workflows via MCP requests.
• Enabled querying execution logs for live runs for up to 30 days per request, looking back up to 60 days.
• Enabled retrieving workflow statistics for enrollments, completions, and failures per step by hour, day, or week.
• Editing a published workflow via MCP automatically sets it to draft, which turns off triggers and pauses active contacts.
• Access requires sub-accounts to have the Workflows API enabled. HighLevel has not confirmed general rollout dates, limits, or plan gating.

Where to find it

HighLevel has not published an in-app navigation path for this update. The feature operates through external AI developer tools (such as Claude, ChatGPT, or Cursor) connected to the LeadConnector MCP server for sub-accounts with the Workflows API enabled.

How agencies will use this

An operator connects an external AI tool like Claude or Cursor to the LeadConnector MCP server for a specific sub-account. Instead of clicking through the HighLevel builder, the operator asks the AI to inspect execution logs over the past 30 days to diagnose why contacts are dropping off. After pinpointing the failed step, the operator prompts the AI to modify the workflow definition, adjust the trigger, and add a branch. Because updating a workflow via MCP places it into draft mode—shutting off triggers and pausing active contacts—the operator reviews the draft definition and issues a publish request once the flow is verified.

Our take

This is a major update for agencies operating at technical scale or building internal AI tooling. Being able to audit, inspect statistics, and generate draft workflows directly from an AI prompt removes repetitive builder tasks. However, it is strictly in private beta right now. Be mindful of the draft behavior: updating any active workflow via MCP immediately halts triggers and pauses active contacts, which can interrupt live automations if done carelessly.