RedHub AI
How the roll-up gate works
How the Prompt Version gate works
This tool grades live voice/chat agents against your registry of approved prompt versions. It reads 4 agents and rolls them into one verdict — the worst, never the average — then a gate that can only make it worse.
The 4 agents
A live prompt whose content hash matches no approved version in the registry is UNKNOWN VERSION however clean the rest of the fleet - and one unknown-version agent forces the whole fleet to SHADOW PROMPTS LIVE, regardless of how many others are IN SYNC.
Why: Two agents are IN SYNC and one is merely DRIFTED to an older approved version - but after-hours-C was edited directly in the console and its content hash matches no approved version at all, an unapproved shadow prompt. Fleet posture is the worst agent, never the average, so it reads SHADOW PROMPTS LIVE.
No single agent is the crisis — the clustering is. The roll-up takes the worst, the gate escalates when trouble compounds, and it names the one thing to fix first. Hashes text you export; it connects to nothing and grades no person. Not legal advice.