# Decide whether a workflow earns reuse Published and revised: October 13, 2026 · Version 1.0.0 Compare a candidate against a simple baseline, inspect failures, and define the upkeep required before relying on it repeatedly. ## Goal and proposed time Reach an evidence-proportional keep, revise, or hold decision about one reusable workflow candidate. Allow 150–210 minutes, including selected reading, making the outputs, and review. This is a proposed study budget, not measured completion time. Split the path across sessions if needed; stop at a checkpoint rather than rushing an evidence check. ## Prerequisites and inputs - Complete the inspectable-pack path or supply an equivalent bounded brief - Be able to save separate candidate versions and review observable outputs - If using a model, record the visible interface and settings without inventing hidden configuration Use only the linked guides and their public or synthetic fixtures. A text editor, calculator, and timer are enough. No provider connection, account change, publication, or outside communication is required. ## Work through the path ### 1. Run one artifact experiment from brief to rollback Read [the guide](/workspace-development/artifact-experiment-evidence-rollback/). Inspect the synthetic source packet and accepted/rejected examples. If you run the optional verifier, record that execution separately from a model comparison. Carry forward: A small artifact contract and a retained known-good copy. ### 2. Compare two prompts with a fair test Read [the guide](/workspace-development/design-controlled-prompt-comparison/). Freeze one meaningful prompt change, six cases, expected behavior, and scoring. Keep two cases held out from prompt revision. Plan repetitions within your study budget. Carry forward: A comparison plan that records conditions, denominators, and unrun cases honestly. ### 3. Turn a useful prompt into a tested workflow Read [the guide](/workspace-development/test-reusable-workflows/). Draft the smallest reusable form, then run and preserve the planned tests if your setup permits. If essential tests are unrun or fail, choose hold or revise rather than calling the candidate tested. Carry forward: A candidate with a clear trigger, inputs, procedure, boundaries, and evidence-proportional disposition. ### 4. Decide what a workspace change should trigger Read [the guide](/workspace-development/triage-change-impact-maintenance/). Name what a source, tool, permission, or audience change would invalidate. Preserve the accepted candidate and define who reviews a repair or retirement. Carry forward: A maintenance owner, review triggers, and rollback limits. ## Worked checkpoint The missing-venue sample in the comparison guide demonstrates a scoring distinction: inventing a town hall fails the factual gate, while preserving “undecided” passes that check. Those authored samples do not establish which prompt will perform better in your actual comparison. The artifact lab’s deterministic checks likewise do not establish model productivity gains. ## Output A frozen comparison plan, candidate and baseline records, explicit test outcomes or unrun labels, and a maintenance note with a known-good copy. ## Exit checks - Both candidates receive equivalent sources, permissions, and cases - Development cases and held-out cases remain distinguishable - Hard factual or boundary failures are not averaged into a passing style score - Observed effort is separated from illustrative arithmetic and unknown cost - A reviewed candidate is not claimed installed, activated, or universally reliable ## Decide what happens next Keep the result when the checks pass. Revise a specific failed check in a new draft, preserving the first attempt. Hold any part that needs missing evidence or authority; an honest hold can complete this learning path even though the wider real-world task remains unfinished. Do not call an authored answer key an observed result. IIEN means infer intent and extrapolate needs here: use it to notice a useful next question and label assumptions. It does not authorize new actions, expand access, or tune model weights. ## Sources and scope This is an editorial sequence of the linked existing guides. Their source sections support the individual methods; the sequence, checkpoints, and study budget are original teaching choices. This path does not claim certification, deployed capability, or measured performance improvement.