Verified

Sevalla MCP Server

Sevalla PaaS API access through sandboxed V8 execution of dynamic tool calls.

Data last scanned today · Reviewed today

23 1 dangerous tool2 tools★ 8TypeScriptMIT

Overview

This server implements the Code Mode MCP pattern to manage Sevalla infrastructure. Instead of creating individual tools for hundreds of API endpoints, it provides two primary tools: 'search' to query the OpenAPI specification and 'execute' to run JavaScript code directly against the Sevalla PaaS API. By leveraging sandboxed V8 isolates, the server enables AI models to discover and invoke API operations on demand. This approach significantly reduces context window overhead by eliminating the need for massive tool definitions, allowing users to perform complex infrastructure management through conversational commands.

Our verdict

This server carries a high-risk score of 38/100. The primary security concerns stem from the fact that the project is community-maintained rather than developed by the official vendor, and critical security aspects like authentication methods and read-only mode support have not yet been editorially reviewed. Given these risks, it is recommended only for users who are comfortable with the security implications of third-party infrastructure integrations.

Reduces context window usage by approximately 99 percent
Uses sandboxed V8 isolates for API execution
Small dependency footprint with only six direct packages
Maintains recent development activity with commits from the last day

Lacks official maintenance by the Sevalla organization
Has not undergone an editorial security or code review
Authentication and read-only modes lack verified security standards

Setup

Claude Desktop

{
  "mcpServers": {
    "sevalla-hosting-mcp": {
      "args": [
        "-y",
        "sevalla-mcp"
      ],
      "command": "npx"
    }
  }
}

Paste into ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows), then restart Claude Desktop. Some servers require API keys or tokens as environment variables — check the project's README and add an "env" object if needed.

No read-only mode — running this grants full read/write access.
No built-in authentication — anyone with access to this config can invoke every tool it exposes.

Tools

Tool	Description	Risk
search	query the OpenAPI spec to discover endpoints, parameters, and schemas	low
execute	run JavaScript in a sandboxed V8 isolate that calls the API via sevalla.request()	high

Compatibility

Client	Local	Docker	Remote	Read-only
Claude Desktop
Cursor
VS Code
Windsurf
ChatGPT

Frequently asked questions

›What tools does this MCP server provide?

The server provides two tools: 'search' for discovering API endpoints, parameters, and schemas, and 'execute' for running JavaScript in a sandboxed V8 isolate to perform API calls.

›How is authentication handled for this server?

Authentication is managed via OAuth. Your MCP client will open a browser to log in with your Sevalla account, and no API keys are required in the configuration file.

›Can I restrict the agent's ability to modify infrastructure?

Yes, you can create a read-only API key on the Sevalla platform if you want to limit the agent to querying infrastructure without making modifications.

›How do I revoke access after uninstalling the server?

You must remove the server configuration from your client and revoke the associated API key at app.sevalla.com/api-keys. Depending on your client, you may also need to manually clear stored OAuth tokens from your system's keychain or the specified local directories.

›What are the environment requirements for this server?

The server requires Node.js version 24 or higher.