Get the FREE Ultimate OpenClaw Setup Guide →

agents

Create AI Agents in a No-Code Visual Builder or TypeScript SDK with full 2-way sync. For shipping AI assistants and multi-agent AI workflows.

Installation
Run this command in your terminal to add the MCP server to Claude Code.
Run in terminal:
Command
claude mcp add --transport stdio inkeep-agents npx -y inkeep-agents-mcp-server \
  --env MCP_API_KEY="your-api-key" \
  --env MCP_API_URL="https://api.inkeep.com" \
  --env MCP_WORKSPACE_ID="your-workspace-id"

How to use

The Inkeep Agents MCP server exposes a centralized platform to manage AI Agents defined in two synchronized formats: a No-Code Visual Builder and a TypeScript SDK. This MCP server orchestrates agent definitions, their sub-agents, and credentials, enabling real-time, two-way synchronization between the visual builder and the SDK codebase. You can deploy Agents to act as real-time chat assistants or as AI-driven workflow automations across channels. Use the MCP APIs to deploy, evaluate, and execute agents, and leverage the built-in observability features to monitor traces and performance across conversations.

To get started, authenticate with your workspace and connect your Visual Builder (agents-ui) or your TypeScript SDK (agents-sdk) to the MCP server. Use the inekeep push/pull utilities to sync between the SDK code and the Visual Builder, ensuring changes propagate in both directions. When constructing agents, you can define subAgents, prompts, and capabilities, then test interactions within the platform. The server supports triggering agents via MCP, A2A, or Vercel SDK APIs, and exposes a REST API to manage configurations, credentials, and projects.

How to install

Prerequisites:

  • Node.js (LTS) installed on your system
  • npm or corepack available
  • Git installed

Installation steps:

  1. Install Node.js and npm from https://nodejs.org/
  2. Install the MCP server package via npx (example):
    • Ensure you have network access to fetch packages
    • Run: npx -y inkeep-agents-mcp-server
  3. Alternatively, install via your package manager if the package is published (replace with actual package name):
    • npm i -g inkeep-agents-mcp-server
  4. Configure environment variables as needed (see env placeholders in mcp_config):
    • Create a .env file or export variables in your shell
  5. Start the MCP server using the command specified in mcp_config (the example uses npx with the package name):
    • npx -y inkeep-agents-mcp-server
  6. Verify the server is running and accessible via the configured MCP API URL.

If you are deploying via Docker or another method, follow your platform’s packaging workflow and ensure the environment variables are provided at runtime.

Additional notes

Tips and common issues:

  • Ensure your API key and workspace ID are correctly set in the environment variables; misconfigurations will prevent authentication.
  • The server supports two-way synchronization between the Visual Builder and the TypeScript SDK; use the provided push/pull utilities to keep both representations in sync.
  • When upgrading the MCP server, review breaking changes in the release notes and update both the agents-sdk and agents-ui integrations accordingly.
  • For large agent configurations, consider enabling observability via OpenTelemetry and the Traces UI to diagnose performance bottlenecks.
  • If you run into network issues fetching packages via npx, verify your network proxy settings and npm registry configuration.
  • You can start with a minimal agent definition to validate the workflow before scaling to full production agents.

Related MCP Servers

Sponsor this space

Reach thousands of developers