Connect agents fast. Manage tasks through MCP only.

How it works

1. Get account

   Sign up once in this app, then log in to get your user session token.

2. Connect your AI app

   Use Codex, Claude Code, or your custom AI app as the MCP client.

3. OAuth and store key

   Click Generate MCP key. Store the returned client_id, client_secret, and refresh token securely in your AI tool.

4. Start sending MCP commands

   Exchange credentials for an access token, then call POST /mcp with JSON-RPC methods like tools/list and tools/call.

Use cases

• Capture todos directly from AI chats and agent workflows.
• Generate suggestions from conversation context and approve them quickly.
• Run daily planning and closeout through MCP tools without opening a dashboard.

Generate MCP OAuth key

After signup/login, generate a managed MCP OAuth key package for your agent. This returns client_id, client_secret, token endpoint, MCP endpoint, and an initial access/refresh token set without custom client setup.

Claude Code OAuth connect

After generating your managed MCP key, run this command in Claude Code. Claude opens browser OAuth to /oauth/authorize; sign in with your notaui email/password and approve access.

Generate MCP key first to get your Claude Code connect command.

OAuth + MCP smoke test

This validates your managed key package by exchanging an OAuth token and calling tools/list on MCP.

Generate MCP key first, then run smoke test.

MCP usage

Use the generated managed credentials with OAuth token exchange:

curl -X POST "$API_BASE/oauth/token" \
  -u "$CLIENT_ID:$CLIENT_SECRET" \
  -d "grant_type=client_credentials" \
  -d "scope=tasks:read tasks:write"

curl -X POST "$API_BASE/mcp" \
  -H "Authorization: Bearer $ACCESS_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"jsonrpc":"2.0","id":1,"method":"tools/list"}'

Session

FAQ