Phase 3 — Define your first goal

Author one business-meaningful goal with explicit success criteria and evidence you would trust in a PR review—before you optimise for CI scale.

Anchor on:

  • Clear user outcome in the goal description (what “green” means for the product).
  • Machine-checkable signals first; rubric/judge steps only where unavoidable.
  • Timeouts and retries realistic for CI, not only your laptop.

Test definition · ← Phase 2 · Phase 4 — Local run →

Agent prompts

Optional: use these with a coding agent to implement this phase in your repo in a safe, gated sequence. Expand only when you need them.

  1. Draft one goal for our highest-risk AI journey; write success criteria as bullet assertions.
  2. Identify minimal browser or API steps the runner must perform to observe those signals.
  3. Peer-review: would a teammate trust this goal in a failed PR without re-running the whole suite blind?