Kangram MCP
Open app
Model Context Protocol

Give your AI agent a task board

The Kangram MCP server exposes task management as tools any MCP-compatible agent can call. Agents stop losing track of work in chat history — tasks land on a real board.

Setup in 60 seconds See supported agents

Setup

One config block. Generate an API key in Kangram, then point your client at the server.

mcp config 
mcp_servers:
  kangram:
    command: npx
    args:
      - kangram-mcp@latest
    env:
      KANGRAM_API_KEY: your-api-key
terminal — agent session 
› Create a task: fix login bug in board 15
✓ Task #142 created · status: Pending

› List all tasks assigned to ivan
✓ 4 tasks: #138 #140 #141 #142

› Move task 42 to Done
✓ Task #42 → Done
  1. 1 Create a board in Kangram (or Telegram).
  2. 2 Generate an API key in Settings → API keys.
  3. 3 Add the config block to your MCP client.
Get your API key

Supported agents

Any MCP-compatible client. A few we test against:

🧠

Claude Code

Create, list, move, and assign tasks from any session.

✦

Cursor

Track editor TODOs on a shared, persistent board.

◈

OpenCode

Hand work off to a backlog across runs.

⬡

Codex

Plan, assign, and route work to teammates.

⚡

Hermes

Stateful agent runs with a real task memory.

🌊

Windsurf

Keep context coherent across long sessions.

Want the full picture?

Read the complete guide to managing tasks with the Model Context Protocol.

Read the MCP guide