Human-in-the-loop security layer. Intercepts high-risk commands and requires push notification approval.
Initial release of ClawShell: Human-in-the-loop security layer for shell command execution. - Intercepts shell commands, analyzes risk, and enforces push notification approval for high-risk operations. - Categorizes commands into Critical (blocked), High (approval required), Medium (logged), and Low (allowed). - Provides three main tools: `clawshell_bash` (secure command exec), `clawshell_status` (view pending/recent approvals), and `clawshell_logs` (audit logs). - Supports push notifications via Pushover or Telegram. - Highly configurable through environment variables and `config.yaml`. - Enforces all decisions via a persistent JSONL audit log.