Show HN: HAL – Harmful Action Limiter: Lean command guard for AI coding agents.github.com/otherland3 pointsotherland263 months ago