Repeat Diagnose → Review → Eval
After run creates a new implementation diff, repeat workbench diagnose, workbench review and workbench eval independently before approving, defining another defect or escalating policy.
Loop evaluation record
From a new diff to the next loop decision
The new diff has its own diagnosis, review and eval, with every claim bounded by its evidence.
Sibling repository../payments-validation-fixture836a75dThe preparation guide explains how to verify a newer revision before using it.
Why this stage exists
A repair is a new change. The original diagnosis and evidence cannot prove that the repair introduced no new defect.
Blocked: the unseen assessment change and hidden ground truth do not exist yet, so independent Stage 7 scoring cannot be delivered.
What you start from
Work from these inputs only. Anything not on this list is either a later stage’s concern or a decision you do not own.
- New implementation diff
- The changed result produced by Stage 6.
- Previous defect specs
- The requirements the run attempted to satisfy.
- Full workbench
- Diagnose, review and eval with no prescribed finding or check.
Send every new diff through the first three commands
One worked pass, so the shape of the work is visible before you do it on your own.
- Run workbench diagnose on the new refund-retry diff.
- Choose review methods and run workbench review against the new risks.
- Run workbench eval and read unexamined or newly silent evidence first.
- Approve, return a new gap to Define, or escalate a policy question.
What this turns onThe loop repeats the first three commands after every run. A green run is only an input to eval.
Do the work
- Check whether the logging repair introduced another card-data sink.
- Review the changed retry implementation and tests for silent regression.
- Evaluate whether the new evidence supports the defect specs.
- If a new defect appears, return to Stage 4; if the diff changes again, repeat this stage.
You produce
Loop evaluation record
- New diff risks
- Review findings
- Eval gaps
- Decision
- Next stage
Challenge your own result
Why run Diagnose again when the repair addressed the original finding?
Show how this resolves
Because the repair can introduce a new card-data path, regression or test weakness. The old diagnosis describes the old diff, not the new one.
Ready to continue when
- The new diff has its own diagnosis, review and eval.
- Any new defect returns to Define instead of being patched ad hoc.
What transfersThe repeatable habit is diagnose, review, eval, define, plan, run—and repeat the first three on every new diff.