Skip to main content

​
What is MCP?

Model Context Protocol (MCP) is an open standard for connecting LLMs to external data sources and tools. Zencoder acts as an MCP client — it connects to MCP servers that expose specific capabilities like database queries, API calls, or browser automation. For basic setup, see Integrations & MCP. This page covers advanced configuration, custom server development, and troubleshooting.

​
Configuration Reference

MCP servers are configured differently depending on your IDE:
Add servers to your VS Code settings under zencoder.mcpServers:
// .vscode/settings.json or User Settings
"zencoder.mcpServers": {
  "server-name": {
    "command": "executable",
    "args": ["arg1", "arg2"],
    "env": {
      "API_KEY": "your-key"
    }
  }
}

​
Configuration Properties

PropertyRequiredDescription
commandYes (stdio)The executable to run (e.g., npx, uvx, docker)
argsNoArray of command-line arguments
envNoEnvironment variables passed to the server process
urlYes (HTTP)URL for remote HTTP-based MCP servers

​
Connection Protocols

Zencoder supports three MCP connection protocols:

​
stdio (Local Servers)

The server runs as a subprocess on your machine. Communication happens over stdin/stdout. 
"zencoder.mcpServers": {
  "filesystem": {
    "command": "npx",
    "args": ["-y", "@modelcontextprotocol/server-filesystem", "/path/to/allowed/dir"]
  }
}

​
Streamable HTTP (Remote Servers)

For servers running on a remote host, use HTTP-based connections: 
"zencoder.mcpServers": {
  "remote-db": {
    "url": "https://mcp.example.com/db"
  }
}

​
OAuth2 (Authenticated Remote Servers)

For third-party services requiring OAuth2 authentication: 
"zencoder.mcpServers": {
  "github-mcp": {
    "url": "https://mcp.github.com",
    "auth": "oauth2"
  }
}
Platform Support:
  • JetBrains: Full support for stdio, HTTP, and OAuth2
  • VS Code: Supports stdio. HTTP and OAuth2 support is being rolled out — check the changelog for status.
These servers are available in the built-in MCP Library (Agent Tools → MCP Library):
ServerWhat it doesInstall
GitHubRepos, PRs, issues, code searchnpx @modelcontextprotocol/server-github
PostgreSQLQuery databases, inspect schemasnpx @modelcontextprotocol/server-postgres
FilesystemRead/write files outside workspacenpx @modelcontextprotocol/server-filesystem
SlackRead channels, send messagesnpx @modelcontextprotocol/server-slack
FigmaExtract design data for code generationVia MCP Library
SentryPull error details and stack tracesVia MCP Library
LinearIssues, projects, and workflowsVia MCP Library
Brave SearchWeb search for documentation and APIsnpx @modelcontextprotocol/server-brave-search
Browse 100+ servers in the MCP Library — open Agent Tools from the menu and click MCP Library.

​
Building a Custom MCP Server

If no existing server fits your needs, you can build your own. MCP servers are lightweight programs that expose tools via the MCP protocol.

​
Quick Start with Python

# server.py
from mcp.server.fastmcp import FastMCP

mcp = FastMCP("my-tool")

@mcp.tool()
def get_deployment_status(service: str) -> str:
    """Check the deployment status of a service."""
    # Your logic here — query your deployment API
    return f"{service} is running v2.3.1 (healthy)"

if __name__ == "__main__":
    mcp.run()
Register it in your MCP configuration: 
"zencoder.mcpServers": {
  "deployments": {
    "command": "python",
    "args": ["path/to/server.py"]
  }
}

​
Quick Start with TypeScript

// server.ts
import { McpServer } from "@modelcontextprotocol/sdk/server/mcp.js";
import { StdioServerTransport } from "@modelcontextprotocol/sdk/server/stdio.js";
import { z } from "zod";

const server = new McpServer({ name: "my-tool", version: "1.0.0" });

server.tool("get_deployment_status",
  { service: z.string() },
  async ({ service }) => ({
    content: [{ type: "text", text: `${service} is running v2.3.1 (healthy)` }]
  })
);

const transport = new StdioServerTransport();
await server.connect(transport);

​
Resources

​
Tool Permissions

Agents request permission before invoking MCP tools. You control this through the permission prompt:
  • Allow once — permit the tool call for this interaction
  • Always allow — trust this tool for all future calls (configurable in settings)
  • Deny — block the tool call
You can configure trusted tools in Zencoder settings to skip the permission prompt for tools you use frequently.

​
Troubleshooting

  • Verify the configuration syntax in your settings
  • Check that the command executable is available in your PATH
  • Restart the IDE after adding a new server
  • Look at the Zencoder output panel for error messages
  • Ensure the server implements the MCP protocol correctly
  • Check that required environment variables are set in the env block
  • For Docker-based servers, verify Docker is running
  • Test the server independently: echo '{"jsonrpc":"2.0","method":"tools/list","id":1}' | your-command
  • OAuth2 is fully supported on JetBrains; VS Code support is being rolled out
  • Clear cached tokens and re-authenticate
  • Verify the OAuth2 redirect URL is configured correctly in the service provider
  • Check server logs for unhandled exceptions
  • Ensure the server handles malformed input gracefully
  • For remote servers, verify network connectivity and firewall rules
  • Increase timeout values if your server processes large datasets

Integrations & MCP

Basic MCP setup and the built-in MCP Library

Coding Agent

The primary agent that invokes MCP tools