Petri 3.0 can run audits against a model's real system promptThe standing instructions a model receives before any user input — defining its role, rules, tools, and tone for the whole conversation.Full definition → and scaffold, reducing test-awareness. It is now maintained by Meridian Labs alongside Inspect, so it can be adopted as a neutral evalA repeatable test for AI quality — a set of tasks plus scoring — used the way software teams use test suites, because model output is too variable to judge by eyeballing.Full definition → tool for agents.
Terms in this piece · Glossary
alignment — The work of making AI systems actually pursue what their builders and users intend, rather than something subtly or dangerously different.
system prompt — The standing instructions a model receives before any user input — defining its role, rules, tools, and tone for the whole conversation.
eval — A repeatable test for AI quality — a set of tasks plus scoring — used the way software teams use test suites, because model output is too variable to judge by eyeballing.
