scrumdo-mcp

Connect any MCP-compatible AI tool (Claude Code, Cursor, Windsurf, and others) directly to your ScrumDo boards.

Once installed, your AI assistant can read cards, move them, create tasks, post comments, and search across your board — without you copy-pasting anything.

Installation

pip install scrumdo-mcp

Cursor quick-start

1 — Install

# To run the server only:
pip install scrumdo-mcp

# To run tests too:
git clone https://github.com/ScrumDoLLC/scrumdo-mcp.git
cd scrumdo-mcp
pip install -e ".[dev]"
pytest tests/ -v

2 — Get your token

Log in to ScrumDo → your org → Settings → API Tokens → Create Token. Copy it — shown once only.

3 — Add to ~/.cursor/mcp.json

{
  "mcpServers": {
    "scrumdo": {
      "command": "scrumdo-mcp",
      "env": {
        "SCRUMDO_TOKEN": "your-token-here",
        "SCRUMDO_ORG": "your-org-slug",
        "SCRUMDO_PROJECT": "your-default-project-slug"
      }
    }
  }
}

Your org and project slugs are the short names in your board URL: app.scrumdo.com/my-company/engineering

4 — Restart Cursor

Done. In any Cursor chat you can now ask:

• "What cards are in the current sprint?"
• "Move ENG-42 to In Review"
• "Add a comment to ENG-42: PR is up for review"
• "List all cards assigned to me"

Setup

Step 1 — Get your token

Log in to ScrumDo → your organization → Settings → API Tokens → Create Token.

Copy the token — it is only shown once. This is your personal key; keep it private.

Step 2 — Configure your AI tool

Find your tool's MCP config file and add the scrumdo server entry:

{
  "mcpServers": {
    "scrumdo": {
      "command": "scrumdo-mcp",
      "env": {
        "SCRUMDO_TOKEN": "your-token-here",
        "SCRUMDO_ORG": "your-org-slug",
        "SCRUMDO_PROJECT": "your-default-project-slug"
      }
    }
  }
}

Your org slug and project slug are the short names in your board URL: app.scrumdo.com/my-company/engineering

Step 3 — Restart your AI tool

Done. Your AI assistant now has direct access to your board.

What you can do

Once connected, just talk to your AI tool naturally:

"What's the status of ENG-42?"
"Move ENG-42 to In Review and add a comment saying the PR is up"
"List all cards assigned to me in the current sprint"
"Create a sub-task on ENG-42: write release notes"
"Search for cards about the login bug"
"What did the team work on this week?"
"Block ENG-42 — waiting on design approval"
"Move ENG-42 to the Sprint 14 iteration"
"Set the due date on ENG-42 to 2026-04-30"
"Assign ENG-42 to Sarah"

Available tools (130 total)

For a governed verification loop, an orchestrator agent calls start_verification_loop (by VerificationProfile slug or inline proof_requirements/verifier_agent), the Maker (Grok/Codex) implements and calls run_verifier against the accepted spec (never self-verifies), and log_loop_step / attach_evidence write the audit trail to the card. When an agent runs inside a loop (SPRYNG_LOOP_ID set), the loop-scoped tools default to that loop, so loop_id is optional.

Environment variables

Token scope

Your API token is restricted to your organization's board data only — cards, tasks, comments, members, iterations. It cannot access billing, account settings, or any other organization's data. Revoke it at any time from Settings → API Tokens.

What is MCP?

Model Context Protocol is an open standard for connecting AI tools to external services. Claude Code, Cursor, Windsurf, and other AI editors support it natively. Install the server once; any MCP-compatible tool can use it.

License

MIT