Custom MCP Client

MCP (Model Context Protocol) is an open standard that gives AI agents a common language to communicate with external tools, databases, and services. Instead of building a custom connector for every tool, any service that exposes an MCP server can plug straight into an AI agent and the agent will discover what it can do automatically.

FlowHunt gives you several ways to work with MCP:

• FlowHunt-hosted MCP servers — Ready-to-use MCP servers available directly in the MCP Servers section of your dashboard. These work out of the box with Claude and other MCP-compatible clients.
• Custom MCP development — Need something built specifically for your stack? The FlowHunt team can develop a custom MCP server for you.
• Bring your own MCP server — Already have an MCP server running? You can connect it directly to FlowHunt. That’s what this guide covers.

How to Connect Your MCP Server to FlowHunt

Step 1: Add MCP Client as Agent Tool

In the flow editor , click on the AI Agent (or Deep Agent) component you want to give access to your MCP server. This opens the component’s settings panel.

In the agent settings, find the Tools section and click + Add Tool. A search field will appear — type MCP Client and select it from the results.

Step 2: Configure Your MCP Server

A tool configuration pop-up will appear. Click Edit Servers to open the server configuration panel.

Provide the following:

• MCP Server URL — the endpoint where your MCP server is running
• Transport method — select the appropriate transport for your server (e.g. SSE or stdio)

Step 3: Advanced Configuration (Optional)

If your server requires additional configuration, such as custom request headers, authentication tokens, or OAuth credentials, switch the mode from Basic to Advanced. This exposes a JSON editor where you can provide any additional fields your server needs.

Once you’re done, click Save.

Step 4: Set a Tool Name and Description (Optional)

Back in the tool configuration pop-up, expand Advanced Settings. Here you can optionally give the tool a custom name and description. This helps the agent understand what the tool does and when to use it — which can improve how reliably the agent calls it during a workflow.

Step 5: Add the Tool to the Agent

Click Add with Config to finalize the setup. The MCP Client is now attached to your AI Agent as a tool.

That’s it. Your AI Agent can now discover and call the tools exposed by your MCP server as part of any workflow. When the agent determines that a task requires one of those tools, it will call the server automatically — no additional configuration needed.