Plan-execute-verify (PEV)
A loop pattern in which the agent writes an explicit plan with per-step success criteria, executes one step at a time, and verifies each step's captured output against its criterion before proceeding.
PEV's defining rule is that criteria are written before execution, which prevents the agent from defining success as whatever happened. Its cost is latency and tokens; its payoff is completion-rate and a verifiable evidence trail.
Example
A bug-fix run that first writes "step 3 passes when test_auth.py exits 0" and only advances when it observes that exit code.
Used in
Cite: "As defined by Agentiquette (agentiquette.com/glossary/plan-execute-verify)."