ReleasesThursday, November 5, 2026
Hermes Skill Lock + Pre-LLM Blocking Hooks (v0.22.6/7 window)
HermesWatcher highlighted key changes from the Nov 1–3 merge window (rolled into v0.22.6/7): Skill Lock lets you lock individual skills from Desktop or CLI so Hermes cannot rewrite them via skill_manage, write_file, patch, or terminal. Plugins and shell hooks can now block a request before it ever reaches the model (no tokens spent). Approval prompts surface hidden Unicode/control characters. `/model reset` restores default model without losing conversation. Cron jobs receive actual run time + timezone.
Why it matters
Directly addresses the #1 complaint of agent self-modification. Gives builders real control over skills and a clean way to enforce guardrails before inference.