Skip to main content

Use Your Agent via MCP Server

MCP makes your agent interoperable. Integrate with tools, editors, or systems that speak the same open protocol.

INFO

TL;DR: Enable MCP in your agent settings, copy the development or production MCP URL, and plug it into your MCP-compatible platform.

Why Enable MCP Support for Your Agent?​

MCP is an open standard that enables AI agents to access context, data, and tools across different systems and environments. When deployed as an MCP server:

  • Your agent can be called by any MCP-enabled app
  • It can maintain multi-turn context across tools
  • It can execute workflows and respond using live data

This is ideal for:

  • AI development teams using workbenches
  • IDE or toolchain integrations
  • Enterprise systems requiring protocol-based interoperability

Prerequisites​

Before connecting your agent to an MCP interface, ensure:

  • Your agent is deployed (test or production)
  • MCP is enabled in Agent Settings
  • You have the correct MCP endpoint URL for your environment
INFO

Reminder: Test environments use the format:
https://<agent-id>.agent.stage.smyth.ai
Production environments use:
https://<agent-id>.agent.psatge.smyth.ai

Step-by-Step: Enable and Integrate MCP​

Step 1: Toggle MCP in Agent Settings​

  1. Open your agent in Studio
  2. Click Settings
  3. Find the MCP section
  4. Toggle MCP on

This will generate unique MCP URLs for test and production.

Step 2: Copy Your MCP Endpoint URL​

After enabling MCP, scroll to the Environment section under Deployments to copy your URLs.

  • Test environment:
    https://<agent-id>.agent.stage.smyth.ai/mcp/sse

  • Production environment:
    https://<agent-id>.agent.psatge.smyth.ai/mcp/sse

Choose based on your integration stage:

  • Use the test URL while developing
  • Switch to the production URL when ready for live deployment
Use the test MCP URL during development, and switch to production (psatge) for stable launches.

Security Considerations for MCP Endpoints​

MCP currently lacks a built-in authentication standard.

WARNING

If your SmythOS agent uses OAuth or other authentication layers, it will reject unauthenticated MCP requests.

To add auth protection, use:

  • A custom proxy layer
  • Third-party middleware

Always secure your production MCP endpoints appropriately.

Common Use Cases for MCP Integration​

  • Connect your agent to VS Code or an AI workbench
  • Integrate with internal AI dashboards or tools
  • Add protocol-level context-awareness to enterprise platforms
Paste your MCP URL into any compatible platform. Your agent will begin responding contextually.

What’s Next?​

With MCP enabled, your SmythOS agent becomes a first-class participant in any system using the Model Context Protocol. From here, you can:

  • View logs and debug via Settings → Deployments → Logs
  • Deploy as LLM to expose it via /chat/completions
  • Deploy as ChatGPT to integrate with GPT Builder

Same agent, multiple endpoints. Maximum reusability.

Enable MCP on your SmythOS agent, grab the appropriate URL, and integrate with your preferred MCP-ready platform.