Adding a field report from my agency setup.

I connected GHL to Claude Desktop (macOS, Electron-based app) via the official MCP endpoint. Auth confirmed, OAuth refreshed, and the GHL REST API is fully operational.

However, the tool manifest consistently fails to load into my active Claude session, meaning zero GHL tools surface despite the connector showing as connected in settings.

Ruled out: Brave Shields (irrelevant on Electron), token expiration (manually refreshed), and API credentials (validated via direct curl calls).

The failure appears isolated to the MCP server's tool manifest delivery layer specifically within Claude Desktop sessions. Would love to connect with anyone on the GHL integrations team who has seen this. I have documented the full diagnostic and am happy to share.

Same need here — using Claude connected via the official MCP server (services.leadconnectorhq.com/mcp). It works great for reading contacts, conversations, and opportunities, but there's no create-task tool for Contacts (only the read-only Get All Tasks exists), even though the REST API supports POST /contacts/:contactId/tasks. Would love to see this added so AI agents can create follow-up tasks based on conversation analysis, not just read existing ones.

This is actually already possible using the API private integration. I used this Github repo using Claude to change it from GPT (how its currently setup) to Claude - https://github.com/mastanley13/GoHighLevel-MCP. In was planning on posting my own public repo about the MCP i created if anyone is interested.