Skip to content

Disposing of model review findings

Answer these formative prompts in your existing work record, then compare your reasoning with the notes or a peer. Each answer should name the evidence that would let another reader reproduce the decision.

Use the public answer guidance after your first attempt.

  1. Before reading a model review, what five fields belong in your first review note?
  2. A candidate changes a guard from active and (owner or staff) to owner or staff and active. Which input separates the two expressions, and what result does the contract require?
  3. A model says, “The active-staff example passes, so the guard is correct.” Is that finding confirmed, refuted or cannot tell? Give the missing case.
  4. A model asks the function to parse the string "false", but the contract accepts booleans only. Which disposition applies and why?
  5. What evidence is sufficient to mark a finding confirmed? What evidence is sufficient to mark it refuted?
  6. You inspected the diff and public tests but cannot determine whether a caller relies on an undocumented status. What do you record under cannot tell, and what check would you request?
  7. The reviewer reports no findings. What claim does that support about the supplied context, and what claim does it not support about the change?
  8. Which person owns acceptance in a repository PR, and what evidence must that person see before merging?
  9. A model suggests a command that edits files outside the change directory. What do you inspect before running it, and when do you stop?
  10. Rewrite “the code should be clean” as one review rule with a source, a location or input, and a checkable expected result.

Transfer to your studio change

Pick one model finding from your own record. Reproduce it with a fresh input or source trace, then write one sentence explaining why the evidence supports confirmed, refuted or cannot tell. Name one requirement the model did not mention and the check you used for it.

Spaced retrieval

R1. Which changed requirement, tests and implementation evidence must be reviewed together when a model proposes a behavior change?

R2. What product decision and acceptance cases are required before mapping an unknown result to a failure?