01
The problem it addresses
AI-assisted development can produce technically plausible output before enough of the surrounding system has been inspected to establish ownership, dependencies, failure paths, or the required proof. PRIDE treats that as a process problem: apparent completion is not accepted as evidence that the correct engineering work occurred.
02
Operating sequence
The stable workflow is Read completely → Understand fully → Plan properly → Execute systematically → Test thoroughly → Document clearly → Review honestly. Proper actions include reading relevant files and traces, searching relevant instances, investigating uncertainty instead of guessing, reasoning across downstream effects, implementing at the owning layer, and verifying the affected behavior.
03
Working metric
PRIDE = Proper Actions ÷ Total Actions, with a working target of 95% or greater proper actions. The metric is a development-process standard, not a claim that the protocol independently guarantees software correctness or eliminates model failure.
Evidence boundary
What this entry supports
The protocol is used in 1DevTeam development. Its effectiveness is treated as an engineering practice claim bounded by observed use, not as a completed comparative research finding.
Related material