{ "markdown": "# WxO Agent MCP\n\n[![npm version](https://img.shields.io/npm/v/wxo-agent-mcp.svg)](https://www.npmjs.com/package/wxo-agent-mcp)\n[![npm downloads](https://img.shields.io/npm/dm/wxo-agent-mcp.svg)](https://www.npmjs.com/package/wxo-agent-mcp)\n[![License](https://img.shields.io/badge/license-Apache--2.0-blue.svg)](LICENSE)\n\n

\n \"wxo-agent-mcp\"\n

\n\n

\n \"Add\n

\n\nSimple MCP (Model Context Protocol) server that **invokes a single Watson Orchestrate agent** remotely. The agent is defined once via environment variables or MCP config.\n\nUse this when you want a lightweight MCP that only chats with one agent—no tool management, no agent listing, no flows. Just `invoke_agent(message)` and `get_agent()`.\n\n**Full documentation:** [DOCUMENTATION.md](DOCUMENTATION.md) (includes [screenshots](DOCUMENTATION.md#testing-in-langflow) for VS Code and Langflow). \n**Publishing:** [PUBLISHING.md](PUBLISHING.md) – npm and MCP Registry procedure.\n\n### Architecture\n\n```\n ┌───────────────────────────────────────┐\n │ Cursor • VS Code • Langflow • etc. │\n │ (MCP clients) │\n └──────────────────┬───────────────────┘\n │ stdio / JSON-RPC\n ▼\n ┌───────────────────────────────────────┐\n │ wxo-agent-mcp │\n │ invoke_agent • get_agent │\n └──────────────────┬───────────────────┘\n │ HTTP / REST\n ▼\n ┌───────────────────────────────────────┐\n │ Watson Orchestrate │\n │ agent + tools + LLM │\n └───────────────────────────────────────┘\n```\n\n## Tools\n\n| Tool | Description |\n| ---------------- | ------------------------------------------------------------------------------------------------------ |\n| **invoke_agent** | Send a message to the configured Watson Orchestrate agent. The agent responds using its tools and LLM. |\n| **get_agent** | Get details of the configured agent (name, description, tools, instructions). |\n\n## Configuration\n\nSet these in `.env` or your MCP client config (e.g. Cursor `mcp.json`):\n\n```env\nWO_API_KEY=your-ibm-cloud-api-key\nWO_INSTANCE_URL=https://your-instance-id.orchestrate.ibm.com\nWO_AGENT_ID=your-agent-id\n# Or WO_AGENT_IDs=id1,id2 (first is used)\n```\n\n## Quick Start\n\n```bash\nnpm install\nnpm run build\ncp .env.example .env\n# Edit .env with your credentials and agent ID\nWO_API_KEY=... WO_INSTANCE_URL=... WO_AGENT_ID=... node dist/index.js\n```\n\n### Verify\n\n```bash\n# Default question (\"Hello, who are you?\")\nWO_API_KEY=... WO_INSTANCE_URL=... WO_AGENT_IDs=... npm run test:verify\n\n# Custom question\nnpm run test:verify -- -ask \"What is the weather in Amsterdam?\"\n```\n\nRuns `get_agent` and `invoke_agent` to confirm connectivity.\n\n### Test in VS Code\n\n1. Open the `wxo-agent-mcp` folder in VS Code.\n2. Run `npm run build`.\n3. `.vscode/mcp.json` registers the MCP server as **wxo-agent** and loads `.env`.\n4. Open Copilot Chat (Ctrl+Shift+I) and ask: _Use invoke_agent to ask: What is the weather in Amsterdam?_ \n For \"what can you do\", use _Use invoke_agent to ask: What can you do?_ to avoid Copilot calling `get_agent` first.\n5. Or run `npm run test:verify` from the terminal (Ctrl+`).\n\nSee [DOCUMENTATION.md](DOCUMENTATION.md#testing-locally-in-vs-code) for more prompts and setup.\n\n**Question examples:** [Full list →](DOCUMENTATION.md#question-examples)\n\n| invoke_agent | get_agent |\n| --------------------------------- | ----------------------------------- |\n| What is the weather in Amsterdam? | Use get_agent to show agent details |\n| Tell me a dad joke | What tools does my agent have? |\n| What time is it in Tokyo? | |\n| What can you help me with? | |\n\n**Langflow:** Add an MCP Tools component, choose STDIO, set Command=`node`, Args=`[\"/path/to/dist/index.js\"]`, and env vars. See [DOCUMENTATION.md](DOCUMENTATION.md#testing-in-langflow).\n\n## MCP Client Configuration\n\n### One-click install (Cursor Directory)\n\n**[Add to Cursor](https://cursor.directory/mcp/wxo-agent-mcp)** — Opens the Cursor Directory page. After adding, set `WO_API_KEY`, `WO_INSTANCE_URL`, and `WO_AGENT_ID` in Cursor MCP settings.\n\n**For [cursor.directory/mcp/new](https://cursor.directory/mcp/new) submission:**\n\n| Field | Value |\n|-------|-------|\n| **Cursor Deep Link** | `cursor://anysphere.cursor-deeplink/mcp/install?name=WxO%20Agent%20MCP&config=eyJjb21tYW5kIjoibnB4IiwiYXJncyI6WyIteSIsInd4by1hZ2VudC1tY3AiXSwiZW52Ijp7IldPX0FQSV9LRVkiOiJ5b3VyLWFwaS1rZXkiLCJXT19JTlNUQU5DRV9VUkwiOiJodHRwczovL3lvdXItaW5zdGFuY2Uub3JjaGVzdHJhdGUuaWJtLmNvbSIsIldPX0FHRU5UX0lEIjoieW91ci1hZ2VudC1pZCJ9fQ==` |\n| **Install instructions** | https://github.com/markusvankempen/wxo-agent-mcp#mcp-client-configuration |\n\n### Cursor (`.cursor/mcp.json`)\n\n```json\n{\n \"mcpServers\": {\n \"wxo-agent\": {\n \"command\": \"npx\",\n \"args\": [\"-y\", \"wxo-agent-mcp\"],\n \"env\": {\n \"WO_API_KEY\": \"your-api-key\",\n \"WO_INSTANCE_URL\": \"https://xxx.orchestrate.ibm.com\",\n \"WO_AGENT_ID\": \"your-agent-id\"\n }\n }\n }\n}\n```\n\n### VS Code Copilot (`.vscode/mcp.json`)\n\nThe server is named **wxo-agent**. Use `envFile` to load `.env`:\n\n```json\n{\n \"servers\": {\n \"wxo-agent\": {\n \"type\": \"stdio\",\n \"command\": \"node\",\n \"args\": [\"${workspaceFolder}/dist/index.js\"],\n \"envFile\": \"${workspaceFolder}/.env\"\n }\n }\n}\n```\n\nFor npx (after publishing), use `\"command\": \"npx\"`, `\"args\": [\"-y\", \"wxo-agent-mcp\"]`, and add `env` with your credentials.\n\n### Antigravity\n\nSame config as Cursor. Copy `examples/antigravity-mcp.json` or add to your MCP config:\n\n```json\n{\n \"mcpServers\": {\n \"wxo-agent\": {\n \"command\": \"npx\",\n \"args\": [\"-y\", \"wxo-agent-mcp\"],\n \"env\": {\n \"WO_API_KEY\": \"your-api-key\",\n \"WO_INSTANCE_URL\": \"https://xxx.orchestrate.ibm.com\",\n \"WO_AGENT_ID\": \"your-agent-id\"\n }\n }\n }\n}\n```\n\n## vs wxo-builder-mcp-server\n\n| | wxo-agent-mcp | wxo-builder-mcp-server |\n| -------- | --------------------------- | ---------------------------------------------------- |\n| Purpose | Invoke one agent | Full dev toolkit (tools, agents, connections, flows) |\n| Agent | Single `WO_AGENT_ID` | Multiple agents, `WO_AGENT_IDs` |\n| Tools | `invoke_agent`, `get_agent` | 30+ tools (list_skills, deploy_tool, etc.) |\n| Use case | Chat with a specific agent | Build and manage Watson Orchestrate resources |\n\n## Links\n\n- **npm:** [wxo-agent-mcp](https://www.npmjs.com/package/wxo-agent-mcp)\n- **MCP Registry:** [io.github.markusvankempen/wxo-agent-mcp](https://registry.modelcontextprotocol.io)\n- **GitHub:** [markusvankempen/wxo-agent-mcp](https://github.com/markusvankempen/wxo-agent-mcp)\n\n## License\n\nApache-2.0\n", "bytes": 7772, "sha": "190ad28efe1d7d11833c79b36a1818955cb8c45462f46f031c35f4f884a7d449", "repo_slug": "markusvankempen/wxo-agent-mcp", "fonte": "repo", "truncated": false, "api": "https://agentalog.com/api/listings/mcp_io_github_markusvankempen_wxo_agent_mcp_da19f31c/readme" }