Pilot Evaluation Checklist

Reference documentation for Pilot Evaluation Checklist in Adaptive Skills.

Use this checklist at the end of a first consumer pilot.

Adoption quality

• The team used a small enough bundle.
• At least one skill was reused more than once.
• The team could explain why each skill was in scope.
• Optional add-ons, such as premortem, were used only when the task justified them.

Operational quality

• Planning or proof quality improved in at least one recurring task.
• Handoffs became clearer or smaller.
• The pilot did not require full-library adoption.

Boundary quality

• Local project rules stayed outside the canon.
• Domain-specific language did not leak into generic skills.
• Claude/Codex differences were handled as operating choices, not forced parity.

Improvement quality

• The pilot identified at least one useful doc or example improvement.
• The pilot identified whether a local wrapper is needed.
• The next step is explicit: expand, adjust, or stop.

Evidence quality

• The pilot report names real tasks, not only abstract examples.
• The report records why each skill was selected.
• The report separates reusable library feedback from consumer-local constraints.