MCPShell provides a secure bridge between LLMs and operating system commands through a configurable YAML-based interface. Developed by Alvaro, it features constraint-based security using CEL expressions to validate parameters before execution, multiple sandboxing options (including macOS sandbox-exec and Linux firejail), template-based command generation, and robust error handling. The implementation supports both direct command execution and server mode, making it particularly valuable for scenarios where AI assistants need controlled access to system utilities, diagnostics tools, or administrative functions without compromising security.
Cognium trust score
50%
Tier
Unverified
Composite of vulnerability cleanliness, spec conformance, provenance, stability, and usage signals — scanned and weighted by Cognium. Human and agent signals are tracked separately.
Returns 7 tools: search_skills, get_skill, list_leaderboard, get_trust_breakdown, resolve_composition, plus the ChatGPT-connector search and fetch. Every tool is annotated read-only.
Resolve this skill directly via MCP tools/call get_skill.