YepCode MCP Server

What is YepCode MCP Server about?

The server bridges AI assistants and YepCode's cloud infrastructure, exposing processes, code execution, environment management, and file storage as MCP tools that AI models can invoke.

How to use YepCode MCP Server?

Obtain an API token from YepCode Cloud (Settings → API credentials).
Choose a deployment method – remote (hosted URL) or local (NPX/Docker).
Configure the MCP client with the server definition (URL or command).
Call MCP tools such as run_code, set_env_var, list_files, or automatically generated run_ycp_<process_slug> from your AI application.

Key features of YepCode MCP Server

Zero‑configuration AI integration – turn any YepCode process into an MCP tool automatically.
Secure code execution in isolated, production‑grade environments.
Environment variable management with optional sensitivity masking.
Full storage API (list, upload, download, delete) for files used by executed code.
Flexible deployment – hosted SSE endpoint, local NPX, or Docker image.
Customizable run_code behavior via YEPCODE_MCP_OPTIONS (disable tool, retain source code).
Debugging support with the MCP Inspector.

Use cases of YepCode MCP Server

AI‑driven assistants that need to run custom scripts or business logic on demand.
Automated workflow orchestration where an LLM selects and triggers YepCode processes.
Dynamic code evaluation in chat‑based coding helpers or education platforms.
Secure file handling for AI models that generate or consume binary assets.
Enterprise integration allowing internal AI tools to interact with existing YepCode pipelines without exposing raw credentials.

FAQ from the YepCode MCP Server

Q: Do I need to install anything locally? A: No, you can use the hosted SSE endpoint. For local use, NPX (npx -y @yepcode/mcp-server) or Docker are provided.

Q: Which environment variable holds the API token? A: YEPCODE_API_TOKEN (or pass it in the Authorization header for HTTP endpoints).

Q: Can I disable the run_code tool? A: Yes, set YEPCODE_MCP_OPTIONS to include disableRunCodeTool.

Q: How are process tools named? A: Each YepCode process tagged with mcp-tool becomes run_ycp_<process_slug> (or <process_id> if the slug is too long).

Q: Where can I debug communication issues? A: Run npm run inspector to launch the MCP Inspector UI.

YepCode MCP Server Preview

What is YepCode MCP Server?

An MCP (Model Context Protocol) server that enables AI platforms to interact with YepCode's infrastructure. Run LLM generated scripts and turn your YepCode processes into powerful tools that AI assistants can use directly.

Why YepCode MCP Server?

Seamless AI Integration: Convert YepCode processes into AI-ready tools with zero configuration
Real-time Process Control: Enable direct interaction between AI systems and your workflows
Enterprise-Grade Security: Execute code in YepCode's isolated, production-ready environments
Universal Compatibility: Integrate with any AI platform supporting the Model Context Protocol

Integration Guide

YepCode MCP server can be integrated with AI platforms like Cursor or Claude Desktop using either a remote approach (we offer a hosted version of the MCP server) or a local approach (NPX or Docker installation is required).

For both approaches, you need to get your YepCode API credentials:

Sign up to YepCode Cloud
Visit Settings > API credentials to create a new API token.

Remote Approach using SSE Server

If your MCP Client doesn't support authentication headers, just use the SSE server URL that includes the API Token. Use a configuration similar to the following:

{
  "mcpServers": {
    "yepcode-mcp-server": {
      "url": "https://cloud.yepcode.io/mcp/sk-c2E....RD/sse"
    }
  }
}

If your MCP Client supports authentication headers, you can use the HTTP server URL that includes the API Token. Use a configuration similar to the following:

{
  "mcpServers": {
    "yepcode-mcp-server": {
      "url": "https://cloud.yepcode.io/mcp/sse",
      "headers": {
        "Authorization": "Bearer <sk-c2E....RD>"
      }
    }
  }
}

Local Approach

Using NPX

Make sure you have Node.js installed (version 18 or higher), and use a configuration similar to the following:

{
  "mcpServers": {
    "yepcode-mcp-server": {
      "command": "npx",
      "args": ["-y", "@yepcode/mcp-server"],
      "env": {
        "YEPCODE_API_TOKEN": "your_api_token_here"
      }
    }
  }
}

Using Docker

Build the container image:

docker build -t yepcode/mcp-server .

Use a configuration similar to the following:

{
  "mcpServers": {
    "yepcode-mcp-server": {
      "command": "docker",
      "args": [
        "run",
        "-d",
        "-e",
        "YEPCODE_API_TOKEN=your_api_token_here",
        "yepcode/mcp-server"
      ]
    }
  }
}

Debugging

Debugging MCP servers can be tricky since they communicate over stdio. To make this easier, we recommend using the MCP Inspector, which you can run with the following command:

npm run inspector

This will start a server where you can access debugging tools directly in your browser.

YepCode MCP Tools Reference

The MCP server provides several tools to interact with YepCode's infrastructure:

Code Execution

run_code

Executes code in YepCode's secure environment.

// Input
{
  code: string;                          // The code to execute
  options?: {
    language?: string;                   // Programming language (default: 'javascript')
    comment?: string;                    // Execution context
    settings?: Record<string, unknown>;  // Runtime settings
  }
}

// Response
{
  returnValue?: unknown;                 // Execution result
  logs?: string[];                       // Console output
  error?: string;                        // Error message if execution failed
}

MCP Options

YepCode MCP server supports the following options:

Disable the run_code tool: In some cases, you may want to disable the run_code tool. For example, if you want to use the MCP server as a provider only for the existing tools in your YepCode account.
Skip the run_code cleanup: By default, run_code processes source code is removed after execution. If you want to keep it for audit purposes, you can use this option.

Options can be passed as a comma-separated list in the YEPCODE_MCP_OPTIONS environment variable or as a query parameter in the MCP server URL.

// SSE server configuration
{
  "mcpServers": {
    "yepcode-mcp-server": {
      "url": "https://cloud.yepcode.io/mcp/sk-c2E....RD/sse?mcpOptions=disableRunCodeTool,runCodeCleanup"
    }
  }
}

// NPX configuration
{
  "mcpServers": {
    "yepcode-mcp-server": {
      "command": "npx",
      "args": ["-y", "@yepcode/mcp-server"],
      "env": {
        "YEPCODE_API_TOKEN": "your_api_token_here",
        "YEPCODE_MCP_OPTIONS": "disableRunCodeTool,runCodeCleanup"
      }
    }
  }
}

Environment Management

set_env_var

Sets an environment variable in the YepCode workspace.

// Input
{
  key: string;                           // Variable name
  value: string;                         // Variable value
  isSensitive?: boolean;                 // Whether to mask the value in logs (default: true)
}

remove_env_var

Removes an environment variable from the YepCode workspace.

// Input
{
  key: string;                           // Name of the variable to remove
}

Storage Management

YepCode provides a built-in storage system that allows you to upload, list, download, and delete files. These files can be accessed from your code executions using the yepcode.storage helper methods.

list_files

Lists all files in your YepCode storage.

// Input
{
  prefix?: string;                       // Optional prefix to filter files
}

// Response
{
  files: Array<{
    filename: string;                    // File name or path
    size: number;                        // File size in bytes
    lastModified: string;                // Last modification date
  }>;
}

upload_file

Uploads a file to YepCode storage.

// Input
{
  filename: string;                      // File path (e.g., 'file.txt' or 'folder/file.txt')
  content: string | {                   // File content
    data: string;                        // Base64 encoded content for binary files
    encoding: "base64";
  };
}

// Response
{
  success: boolean;                      // Upload success status
  filename: string;                      // Uploaded file path
}

download_file

Downloads a file from YepCode storage.

// Input
{
  filename: string;                      // File path to download
}

// Response
{
  filename: string;                      // File path
  content: string;                       // File content (base64 for binary files)
  encoding?: string;                     // Encoding type if binary
}

delete_file

Deletes a file from YepCode storage.

// Input
{
  filename: string;                      // File path to delete
}

// Response
{
  success: boolean;                      // Deletion success status
  filename: string;                      // Deleted file path
}

Process Execution

The MCP server can expose your YepCode Processes as individual MCP tools, making them directly accessible to AI assistants. This feature is enabled by just adding the mcp-tool tag to your process (see our docs to learn more about process tags).

There will be a tool for each exposed process: run_ycp_<process_slug> (or run_ycp_<process_id> if tool name is longer than 60 characters).

run_ycp_<process_slug>

// Input
{
  parameters?: any;                      // This should match the input parameters specified in the process
  options?: {
    tag?: string;                        // Process version to execute
    comment?: string;                    // Execution context
  };
  synchronousExecution?: boolean;        // Whether to wait for completion (default: true)
}

// Response (synchronous execution)
{
  executionId: string;                   // Unique execution identifier
  logs: string[];                        // Process execution logs
  returnValue?: unknown;                 // Process output
  error?: string;                        // Error message if execution failed
}

// Response (asynchronous execution)
{
  executionId: string;                   // Unique execution identifier
}

get_execution

Retrieves the result of a process execution.

// Input
{
  executionId: string;                   // ID of the execution to retrieve
}

// Response
{
  executionId: string;                   // Unique execution identifier
  logs: string[];                        // Process execution logs
  returnValue?: unknown;                 // Process output
  error?: string;                        // Error message if execution failed
}

License

This project is licensed under the MIT License - see the LICENSE file for details.