taskr-mcp-client

stdio to HTTP bridge for Taskr MCP server. Connects AI agents (Claude Code, Cursor, Windsurf, Claude Desktop, OpenClaw) to Taskr's MCP endpoint.

Quick Setup (Recommended)

Add to your .mcp.json or IDE MCP config — no API keys needed:

{
  "mcpServers": {
    "taskr": {
      "command": "npx",
      "args": ["-y", "taskr-mcp-client@latest"]
    }
  }
}

On first use, a browser window opens for sign-in. Credentials are cached at ~/.taskr/credentials.json so you only sign in once.

Claude Code (CLI)

claude mcp add --transport stdio --scope user taskr -- npx -y taskr-mcp-client@latest

Manual Setup (Advanced)

If you prefer to set credentials explicitly:

{
  "mcpServers": {
    "taskr": {
      "command": "npx",
      "args": ["-y", "taskr-mcp-client@latest"],
      "env": {
        "MCP_API_URL": "https://www.taskr.one/api/mcp",
        "MCP_PROJECT_ID": "your-project-id",
        "MCP_USER_API_KEY": "your-api-key"
      }
    }
  }
}

When all three environment variables are set, the OAuth flow is skipped entirely.

CLI Commands

npx taskr-mcp-client auth      # Sign in via browser
npx taskr-mcp-client logout    # Remove cached credentials
npx taskr-mcp-client status    # Show current auth state
npx taskr-mcp-client --help    # Show usage

Credential Resolution Order

1. Environment variables — MCP_API_URL, MCP_PROJECT_ID, MCP_USER_API_KEY (if all set, used directly)
2. Cached credentials — ~/.taskr/credentials.json (from previous OAuth sign-in)
3. Device auth flow — Opens browser for sign-in, polls for authorization, caches result

How It Works

This bridge translates between:

• stdin/stdout (MCP protocol used by AI agents)
• HTTP POST (JSON-RPC 2.0 endpoint used by Taskr server)

It allows any MCP-compatible AI agent to communicate with Taskr's serverless MCP endpoint.

License

MIT