Adaptation Checklist

Using This Project Structure in Any Domain

Use this checklist before starting a new project and before trusting any AI output.
If you cannot check an item, the project is not ready.

1. Decision Classification

☐ Have I identified what kind of problem this is?

☐ Does the classification match the consequences of being wrong?

☐ If the outcome may be defended, challenged, or acted upon, have I treated it as decision-grade?


2. Objective & Scope

☐ Is the decision or question stated in one clear paragraph?

☐ Can I explain what problem this is solving without referencing AI output?

☐ Is the intended use of the outcome explicitly stated?

☐ Are time horizon and scope boundaries defined?


3. Constraints & Non-Goals

☐ Are constraints explicit rather than implied?

☐ Are non-goals stated clearly enough to prevent drift?

☐ Have I listed assumptions the AI must not make?

☐ Is there a change-control rule for modifying constraints?


4. File Authority

☐ Are files explicitly labeled as:

☐ Is it clear which files govern reasoning?

☐ Is there an instruction preventing reference material from silently becoming authority?


5. Evidence Standards

☐ Have I defined what counts as a “fact” in this domain?

☐ Are acceptable source types specified?

☐ Is there a requirement to label inference and speculation?

☐ Are unsourced claims prohibited from driving conclusions?


6. Assumption Control

☐ Is there an active Assumptions Log?

☐ Are high-impact assumptions identified and evaluated?

☐ Has the question “What would make this wrong?” been asked?

☐ Are unresolved assumptions acknowledged?


7. Audit & Challenge

☐ Is there a reusable audit file?

☐ Are alternative interpretations explicitly required?

☐ Is premature closure actively checked?

☐ Can a skeptical reviewer follow the reasoning without chat context?


8. Confidence Discipline

☐ Is confidence explicitly stated for any conclusion?

☐ Is confidence proportional to evidence and uncertainty?

☐ Is at least one factor identified that could change the conclusion?

☐ Is low confidence treated as acceptable where warranted?


9. Completion Criteria

☐ Is it clear what “done” means for this project?

☐ Is deferral an allowed outcome?

☐ Will the decision be recorded in a Decision Summary file?

☐ Can the outcome be defended without appealing to AI authority?


Final Gate

If any section above is unchecked, do not accept conclusions.

Fix the structure first.
Then resume analysis.


Closing Note

This checklist is not about control for its own sake.
It exists to ensure that AI supports judgment rather than replaces it.

If the checklist feels excessive, the problem is likely under-defined—or the risk of being wrong is being underestimated.