ShellGuard
Empowers LLM agents with secure, read-only bash access to remote servers over SSH.
ShellGuard is an MCP server designed to bridge the gap between AI agents and live server environments. It allows Large Language Models to securely SSH into production, staging, or development servers, enabling them to perform tasks like running diagnostics, inspecting logs, querying databases, and troubleshooting issues without manual intervention. The system ensures safety by restricting commands to a curated set of read-only tools, providing intelligent feedback and suggestions when destructive operations are attempted, allowing the LLM to self-correct and continue its investigation.
Key Features
01Provides LLM agents with read-only bash access to remote servers via SSH.
02Enforces a robust command validation pipeline with actionable suggestions for blocked operations.
03Supports both native Go SSH and system OpenSSH for comprehensive `~/.ssh/config` support.
04Enables on-demand provisioning of essential diagnostic tools like `ripgrep`, `jq`, and `yq` on remote hosts.
05Integrates seamlessly with various MCP clients including Cursor, Claude, OpenCode, VS Code, Zed, and Roo Code.
069 GitHub stars
Use Cases
01Automated server diagnostics and health checks by AI agents.
02LLM-driven inspection of logs and troubleshooting of application issues.
03Secure exploration of remote file systems and database queries by AI.