Security middleware for AI Agents. Intercepts shell commands before execution using a multi-layer pipeline: binary allowlist, regex patterns, deterministic intent coherence mapping, and LLM semantic check as last resort.

A hardened local code runner with predictable limits, clean UX, and an agent‑friendly API for executing snippets without heavy infrastructure.

Results utilities and wrappers made for JS

A secure MCP server for executing whitelisted development commands with comprehensive security controls.