SEBBI.PRO · Bias & Drift Test · Deep Check · Trust Centre
DEEP CHECKThe sebbi rules engine catches known prompt attacks in microseconds at zero token cost. Deep Check adds a second opinion from your own AI for the subtle ones: disguised instructions, social engineering, policy breaches written politely. The stricter verdict wins, and only fingerprints are sealed.
Instruction override, prompt extraction, role hijack, data exfiltration, hidden markers, payment push, encoded payloads, urgency pressure. Free and instant.
Claude or OpenAI reads the text against your policy and explains what it finds. Your key, your model, your bill.
Both verdicts and the final call are sealed with fingerprints of the text and policy. The text itself is never stored.
Rules check free. Deep Check: 7 days free, then 50p per device per month. Need a key? Start here.