Independent use

Prove that the researcher can repeat the scenario without consultant help.

Result

The researcher completes a comparable task, verifies the result, and recovers from a representative failure without the consultant choosing the prompt, tool, correction, or acceptance decision for them.

Independence does not mean zero questions. A bounded clarification is acceptable when the researcher still controls the workflow and decision. Consultant takeover is not evidence of independent use.

Actions

1. Choose a new example with comparable inputs, decisions, risks, and output format; do not repeat the guided case unchanged.
2. Give the researcher the accepted guide and project package, then observe without supplying step-by-step prompts.
3. Record the actions, decisions, clarifying questions, deviations, and assistance provided.
4. Introduce or observe one representative failure and let the researcher diagnose the failed layer.
5. Have the researcher correct the project instruction or workflow, then re-run the affected acceptance criterion.
6. Ask the researcher to explain the evidence, limitations, and acceptance decision in their own words.
7. Record the remaining support need and decide accepted, needs another run, or not yet independent.

Check

The record contains one accepted guided run and one accepted independent run on new material. The second run shows no consultant takeover, includes a successful check and recovery path, and produces its own evidence package and explicit decision.

If it fails

Classify the exact capability or instruction that failed and return only to that stage. Do not rename a consultant-completed run as independent. Improve the guide or project rule, then repeat the failed portion on another comparable example.

Next: Handover package.