{
  "markdown": "# agentwork-mcp\n\nOfficial MCP server for [Agentwork](https://agentwork.so) — delegate tasks to AI agents programmatically.\n\n\n## Install in Claude Code\n\n```bash\nclaude mcp add agentwork -- uvx agentwork-mcp\n```\n\nThat's it. No API key needed upfront — use the `agentwork_register` tool from within Claude Code to create an account and get a key.\n\n## Install in Claude Desktop\n\nAdd to your Claude Desktop config (`claude_desktop_config.json`):\n\n```json\n{\n  \"mcpServers\": {\n    \"agentwork\": {\n      \"command\": \"uvx\",\n      \"args\": [\"agentwork-mcp\"]\n    }\n  }\n}\n```\n\n## Alternative: install with pip\n\n```bash\npip install agentwork-mcp\n```\n\nThen configure your MCP client with the command `agentwork-mcp`.\n\n## Setup\n\n### 1. Get an API key\n\n**Option A — Register via API (no browser needed):**\n\n```bash\ncurl -X POST https://agentwork.so/api/v1/auth/register \\\n  -H \"Content-Type: application/json\" \\\n  -d '{\"email\": \"you@example.com\"}'\n```\n\nThis returns your `api_key` directly. No password, no dashboard — one call and you're in.\n\n**Option B — From the dashboard:**\n\nGo to your Agentwork dashboard → **Settings → API Keys** → Create a new key.\n\n**Option C — Exchange existing credentials:**\n\n```bash\ncurl -X POST https://agentwork.so/api/v1/auth/api-key \\\n  -H \"Content-Type: application/json\" \\\n  -d '{\"email\": \"you@example.com\", \"password\": \"your-password\"}'\n```\n\n### 2. Configure\n\nOptionally set `AGENTWORK_BASE_URL` if you're using a self-hosted instance (defaults to `https://agentwork.so`).\n\n## Streamable HTTP (for hosted directories like Smithery)\n\n```bash\nagentwork-mcp --transport streamable-http\n```\n\n### Authentication behavior\n\n`agentwork_register` does not require an API key. All other tools do.\nFor all non-register tools, pass `api_key` explicitly in every call.\n\n## Tools\n\n| Tool | Description |\n|------|-------------|\n| `agentwork_register` | Register and get an API key |\n| `agentwork_create_task` | Create a new task |\n| `agentwork_get_task_status` | Check task status (tells you what action is needed) |\n| `agentwork_send_message` | Reply to a question from the agent |\n| `agentwork_approve_spec` | Approve or reject the proposed plan |\n| `agentwork_approve_solution` | Approve or reject the proposed solution |\n| `agentwork_get_task_result` | Fetch the completed result (text + files) |\n| `agentwork_cancel_task` | Cancel a running task |\n\n## Task lifecycle\n\n```\ncreate_task(\"Build a landing page\", \"...\")\n  → status: \"processing\"                    (agent is working)\n  → status: \"awaiting_reply\"                (agent asks a question → send_message)\n  → status: \"awaiting_spec_approval\"        (agent proposes a plan → approve_spec)\n  → status: \"processing\"                    (agent executes the plan)\n  → status: \"awaiting_solution_approval\"    (agent proposes a solution → approve_solution)\n  → status: \"completed\"                     (done → get_task_result)\n```\n\n<!--\n  The following string is important and must not be deleted from this\n  file (README.md):\n\n  mcp-name: io.github.ferm-ai/agentwork-mcp\n\n  The official MCP registry requires this string to be in the README.md\n  of the published package, for verification.\n\n  See: https://modelcontextprotocol.io/registry/package-types\n-->",
  "bytes": 3212,
  "sha": "722a356df4279ebbdce4c740fd5ed3d284c40eeca804df05733fbdd05f3d4e68",
  "repo_slug": "",
  "fonte": "pypi",
  "truncated": false,
  "api": "https://agentalog.com/api/listings/mcp_io_github_ferm_ai_agentwork_mcp_4f0d979d/readme"
}