claude-desktop-skills-proxy
Verified Safeby CodefiLabs
Overview
Enables Claude Desktop to perform network requests, execute host CLI commands, and manage files by bypassing its sandbox restrictions.
Installation
npx -y github:CodefiLabs/claude-desktop-skills-proxySecurity Notes
The server demonstrates robust security considerations. It explicitly blocks common SSRF targets (localhost, private IPs, AWS metadata endpoint) and dangerous CLI commands (rm, sudo, chmod, package managers like apt/brew). It uses `spawn()` instead of `exec()` for CLI execution, preventing shell injection, and validates arguments against dangerous shell operators. Sensitive file paths (e.g., ~/.ssh, /etc/shadow) are blocklisted for reading and serving. An explicit approval flow is required for new domains and commands, adding a layer of user consent. Rate limiting is implemented to prevent abuse. The main external dependency, `cloudflared` for public tunnels, is checked for installation, but its inherent security is external to this codebase. Overall, the project prioritizes security given its function of extending system access.
Similar Servers
claude-code-mcp
Acts as an MCP server to enable LLMs to run Claude Code CLI in one-shot mode, bypassing permissions for complex coding, file system, Git, and terminal operations.
consult-llm-mcp
An MCP server that allows AI agents like Claude Code to consult stronger, more capable AI models (e.g., GPT-5.2, Gemini 3.0 Pro) for complex code analysis, debugging, and architectural advice.
mcp-client-server
A specialized Model Context Protocol (MCP) server designed to act as a testing harness for developing and iterating on other MCP servers within a Claude AI environment.
converse
Orchestrates and exposes various AI tools (chat, multi-model consensus, job management) over the Model Context Protocol, enabling local, persistent, and potentially asynchronous AI interactions across multiple Large Language Model (LLM) providers.