Bash Security Validator — Production OpenClaw Shell Safety

A 23-validator production security chain for OpenClaw bash execution — sourced directly from Anthropic's Claude Code bash security implementation. Covers text-level obfuscation detection, structural substitution analysis, Unicode/encoding attack prevention, shell-specific dangerous builtin blocking (Bash + Zsh separately), and persistence/escalation vector detection. Every validator Claude Code itself runs before executing shell commands. Installs as a SKILL.md that activates the full chain on every exec tool call.

Want to understand the threat model first? Read the free primer: OpenClaw Bash Safety — covers what ClawHavoc exploited and why regex alone fails.