Your AI assistant can work in minddy.

Ask Claude Code, Codex or Cursor to read your issues, create a plan and update progress. MCP is the connection that lets your assistant use minddy.

Connect your AI assistant

MCP server address

https://www.minddy.app/api/mcp

Approve access in your browser

Available on every minddy plan, including Free. Sign in with your minddy account; no API key is needed.

Connect your AI assistant

Find your assistant below and follow its setup instructions. Then sign in to minddy in your browser to approve access.

  • Let your assistant set it up

    Paste this prompt into your assistant to guide it through connecting to minddy.

    Add the minddy MCP server to my setup. It is a remote MCP server at https://www.minddy.app/api/mcp, over HTTP with OAuth, and there is no API key to create. Read https://www.minddy.app/llms.txt for the exact steps, register it for the agent you are running in, then tell me how to approve access.

  • Claude Code

    Paste this command in your terminal, then authorize access in your browser.

    claude mcp add --scope user --transport http minddy https://www.minddy.app/api/mcp

  • Claude Desktop

    In Claude Desktop or claude.ai: Settings → Connectors → Add custom connector, then paste this URL.

    https://www.minddy.app/api/mcp

  • Codex

    Paste this command in your terminal, then authorize access in your browser.

    codex mcp add minddy --url https://www.minddy.app/api/mcp && codex mcp login minddy

  • Cursor

    Paste into ~/.cursor/mcp.json, then sign in when Cursor asks.

    {
  "mcpServers": {
    "minddy": {
      "url": "https://www.minddy.app/api/mcp"
    }
  }
}

  • Gemini CLI

    Paste this command in your terminal, then authorize access in your browser.

    gemini mcp add --scope user --transport http minddy https://www.minddy.app/api/mcp

  • OpenCode

    Paste into opencode.json (or the global one), then run `opencode mcp auth minddy`.

    {
  "$schema": "https://opencode.ai/config.json",
  "mcp": {
    "minddy": {
      "type": "remote",
      "url": "https://www.minddy.app/api/mcp",
      "enabled": true
    }
  }
}

  • VS Code

    Paste this command in your terminal, then authorize access in your browser.

    code --add-mcp '{"name":"minddy","type":"http","url":"https://www.minddy.app/api/mcp"}'

  • Windsurf

    Paste into ~/.codeium/windsurf/mcp_config.json (Windsurf → Settings → MCP), then sign in when prompted.

    {
  "mcpServers": {
    "minddy": {
      "serverUrl": "https://www.minddy.app/api/mcp"
    }
  }
}

What an agent does in minddy

Your assistant can read and update project information in minddy, so your team can see the results.

  • Copy a prompt

    Each issue gives your agent a prompt to write a plan, implement it or verify the result.

  • The agent works

    It reads the issue over MCP, checks off its plan and opens a pull request.

  • Merge, and minddy follows

    The issue moves to In review, then to Done when the pull request is merged.

  • Reviews on the issue

    Reviews, checks and comments from the pull request show on the issue.

  • New comments

    Comments your agent hasn’t answered show on the card, ready to copy as a prompt.

  • API and webhooks

    Your app can create issues and feedback through the API, and receive webhooks.

  • Read an issue with its full context.
  • Add a task plan and mark completed steps.
  • Update an issue’s status, priority, assignee, or objective.
  • Create issues and objectives, and link them together.
  • Report back in a comment.
  • Read and write the project's wiki pages.
  • Review the pull request opened on the issue.
  • Work with your notes, personal schedule, and user feedback.

Three ways to move work forward

Here are three requests to try after connecting your assistant. Replace MIN-42 with one of your issue numbers.

  • Write an implementation plan

    The assistant reads the issue and saves a step-by-step plan in minddy. If it also has access to your code, it can use that context.

    Ask this

    Read MIN-42 in minddy and write its implementation plan.

  • Keep work up to date

    Ask the assistant to mark each plan task as complete when it finishes. Your team can follow progress directly on the issue.

    Ask this

    Work through the plan on MIN-42 and keep its tasks up to date.

  • Turn decisions into issues

    Describe the work you agreed on. The assistant creates issues with a description and priority, ready for your team to review.

    Ask this

    Create issues in minddy for everything we just decided.

Download minddy

You stay in control

You approve each connection and can remove its access from your account.

  • Your account permissions apply

    The assistant can access the same projects as your account and makes changes on your behalf.

  • One approval, in the browser

    When you first connect, choose your minddy account and approve the access request in your browser.

  • Remove access from your account

    Open the connected apps in your account settings to revoke an assistant’s access.

Questions and answers

Create a project and describe your first issues. Smart Fill and Smart Assign complete them as soon as they are created.

Download minddy