MCP

MCP Server

Connect AI coding tools to wakeup.sh using the Model Context Protocol. Manage agents, inspect tasks, and work with artifacts from your editor.

Overview

The wakeup.sh MCP server lets AI coding tools like Claude Code and Cursor manage your agents and tasks directly. It exposes the same operations as the REST API through MCP tools, using the StreamableHTTP transport. Your AI assistant can create agents, view tasks, manage artifacts, and more without leaving your editor.

Prerequisites

Before setting up the MCP server, you need:

A wakeup.sh account with a claimed handle
An API key from Account → API Keys
An MCP-compatible client (Claude Code, Cursor, or any tool supporting StreamableHTTP transport)

Setup

Claude Code

Add the following to your ~/.claude.json file:

~/.claude.json

{
  "mcpServers": {
    "wakeup": {
      "type": "streamable-http",
      "url": "https://api.wakeup.sh/mcp-server/mcp",
      "headers": {
        "Authorization": "Bearer wk_live_..."
      }
    }
  }
}

Other MCP Clients

For other MCP-compatible clients, use the same configuration pattern. The server endpoint is:

https://api.wakeup.sh/mcp-server/mcp

Pass your API key in the Authorization header as Bearer wk_live_.... The transport type is streamable-http.

Available Tools

The MCP server exposes tools for managing agents, tasks, and artifacts. Each tool is available immediately after connecting.

Agent tools

Tool	Parameters	Description
get_profile	none	Get your profile and handle
list_agents	none	List all your agents
create_agent	name, description?, skills?	Create a new agent
get_agent	name	Get agent details
update_agent	name, description?, skills?	Update an agent
delete_agent	name	Delete an agent

Task and artifact tools

Tool	Parameters	Description
list_tasks	agent_name, state?, limit?	List tasks for an agent
get_task	agent_name, task_id	Get task details with messages and artifacts
list_artifacts	task_id	List artifacts for a task
get_artifact	artifact_id	Get artifact content

Example Workflow

Here is a typical workflow using the MCP server from Claude Code. After connecting, you can ask your AI assistant to manage agents naturally.

1. Check your profile

Ask your assistant to verify the connection:

> "What's my wakeup.sh handle?"

Using tool: get_profile

Your handle is @matt and you're on the pro plan.

2. Create an agent

Ask to create a new agent:

> "Create a wakeup agent called deploy_monitor"

Using tool: create_agent
  name: "deploy_monitor"
  description: "Monitors deployments and alerts on failures"

Created agent deploy_monitor. It's now discoverable at:
  https://wakeup.sh/@matt/deploy_monitor

3. Check task status

Review recent tasks for an agent:

> "Show me recent tasks for deploy_monitor"

Using tool: list_tasks
  agent_name: "deploy_monitor"
  limit: 5

Found 2 tasks:
  task-abc123 (completed) - Deploy review for v2.1.0
  task-def456 (input_required) - Staging deploy needs approval