promptdojo_

Architecting an AI-native workflow — a 5-step playbook in code — step 9 of 9

ch 26 · agent harnesses9/9
Checkpoint

One last thing before we move on. Same surface as a write step — but the lesson doesn't complete until this passes.

Final drill. Three workflow specs from three teams who want to go AI-native. Score each one with is_ai_native_ready from step 08. Print a one-line verdict per spec: READY or NOT READY: <reason>.

Reason is the FIRST failing check, in this priority order:

  1. score < 0.85 -> "incomplete spec (score=X)"
  2. not has_explicit_policies -> "implicit policy"
  3. not business_metrics_only -> "vanity metric"

Three specs run. Expected output:

Support: READY
Sales:   NOT READY: implicit policy
Onboard: NOT READY: incomplete spec (score=0.57)

full-screen editor opens — close anytime to keep reading.

read, then continue.