01Who this is for
- Founders of SaaS startups who want to add AI driven task creation without leaving HighLevel.
- Marketing agency managers who need to update client project statuses through chat commands.
- Operations directors of service firms who prefer natural language prompts to organize Lists and Spaces.
02Project Management MCP, explained
The MCP integration bridges a supported AI client and HighLevel Project Management. Once an OAuth connection is authorized, the AI assistant can discover and invoke supported operations such as creating tasks, assigning custom statuses, updating fields and reading or posting task comments. All actions are performed inside HighLevel, keeping project data in one place.
The connection is established by adding a hosted MCP endpoint in the AI client and completing the OAuth flow. After authorization, the AI client appears as a connected app and can receive natural language requests that translate into HighLevel API calls. No local server, VPN or private integration token is needed; the hosted endpoint handles communication.
Supported AI clients include ChatGPT, Claude.ai, Codex (IDE extension or CLI), Claude Code and Muse. Each client has its own steps to add the MCP server URL and complete OAuth, after which the assistant can manage tasks, subtasks, Spaces, Lists, custom statuses, custom fields and comments.
03How businesses use Project Management MCP
- Marketing agency. The agency enables its copywriters to ask the AI assistant to create new tasks in a client campaign List, set custom status values and attach custom fields. The AI client receives the request, calls HighLevel via MCP and the tasks appear without the copywriter opening the Project Management interface.
- Software development firm. Developers use the Codex IDE extension to add the MCP server, authenticate, then type a natural language prompt to add a subtask under an existing development ticket. The assistant writes the subtask, assigns it, and the update shows right away in HighLevel.
- Event planning business. Event coordinators connect Muse, then ask the AI to read comments on a venue booking task and to add a new comment with the latest vendor confirmation. The assistant retrieves and posts the comment through the MCP connection, keeping the discussion inside HighLevel.
04The numbers and rules that matter
- Supported AI Clients: Claude.AI, ChatGPT, Cursor, Codex, Claude Code, Muse.
- Claude.AI endpoint: Https://services.leadconnectorhq.com/MCP/anthropic/v2.
- ChatGPT endpoint: Https://services.leadconnectorhq.com/MCP/openai/v2/.
- Codex endpoint: Https://services.leadconnectorhq.com/MCP/openai/v2/.
- Muse endpoint: Https://services.leadconnectorhq.com/MCP/muse/v2/.
05Before you start
- A HighLevel subaccount you own or manage.
- An account with the chosen AI client that permits adding custom connectors or apps.
- If you are an agency, the installation must support multi subaccount selection for the OAuth step.
06How to set up and use Project Management MCP in HighLevel
Connect ChatGPT
- Open ChatGPT on the web.
- Open Settings.
- Select Apps.
- If your account requires Developer mode, enable it before creating the MCP connection.
- Click Create to start a custom MCP app.
- Enter a name such as HighLevel Project Management.
- Provide a description that identifies the connection as your HighLevel MCP integration.
- Enter the MCP server URL: https://services.leadconnectorhq.com/mcp/openai/v2/
- Continue with the MCP app configuration.
- Select the available option to scan or discover the MCP tools.
- Begin OAuth authorization when prompted.
- After completing authorization, return to ChatGPT and finish creating the app.
Connect Claude.ai
- Open Claude.ai.
- Select Settings.
- Select Connectors.
- Click Add custom connector.
- Enter the server URL: https://services.leadconnectorhq.com/mcp/anthropic/v2
- Click Connect.
- Complete the HighLevel OAuth authorization process.
- Start a new Claude conversation after the connection is established.
Connect Codex (IDE extension)
- Open the Codex IDE extension.
- Select the gear menu.
- Select MCP servers.
- Click Add server.
- Choose Streamable HTTP.
- Enter the server URL: https://services.leadconnectorhq.com/mcp/openai/v2/
- Select Authenticate.
- Complete the HighLevel OAuth authorization process.
Connect Codex (CLI)
- Add the MCP server with: codex mcp add leadconnector --url https://services.leadconnectorhq.com/mcp/openai/v2/
- Run the login command: codex mcp login leadconnector
- Follow the sign in and authorization prompts.
Connect Claude Code
- Open a terminal.
- Add the hosted MCP server using: claude mcp add --transport http leadconnector https://services.leadconnectorhq.com/mcp/anthropic/v2
- Complete the HighLevel OAuth authorization process when the browser opens.
- Verify the connection with: claude mcp list
Connect Muse
- Open Muse.
- Locate the app or MCP connection options.
- Select the HighLevel app.
- Begin the connection process.
- Complete the HighLevel OAuth authorization flow.
- Return to Muse and confirm that HighLevel is connected.
07Settings that matter
- Developer mode. Enables the creation of custom MCP apps in ChatGPT when the account requires it.
- MCP server URL. Specifies the hosted endpoint that the AI client uses to communicate with HighLevel Project Management.
- OAuth scopes. Define which HighLevel resources the AI connection may access after authorization.
08Things to know before you rely on it
- Agency multi subaccount connections are only available where the installation option exists and may not appear for every account.
- No Private Integration Token is required; the integration relies solely on OAuth.
- Supported AI clients are limited to Claude.ai, ChatGPT, Codex, Claude Code and Muse.
09How it connects to the rest of HighLevel
- Learn how to manage tasks and subtasks inside HighLevel. Task management guide
- Get started with HighLevel Project Management fundamentals. Project management intro
- Review HighLevel pricing and plan options. Pricing page
Michael's take

Project Management MCP feels like a small convenience that can quickly become a real shift for teams that already live inside AI chat environments. The ability to tell your assistant to spin up tasks, set Custom Fields, or pull a comment without opening the HighLevel UI removes a lot of friction, especially when you’re juggling multiple projects at once.
It fits a marketing agency that runs client onboarding workflows across many subaccounts. Imagine the project manager asking the AI to create three onboarding tasks for a new client and assign them. All in one sentence.