Bug

Anchor debugging in a reproducible difference between observed and expected behavior.

Purpose

Anchor debugging in a reproducible difference between observed and expected behavior.

Recommended

  • Include reproduction steps, environment or version, symptoms, and expected behavior.
  • Ask for root cause before selecting a fix.
  • Require a regression test when practical.

Avoid

  • Assuming the first suspicious line is the cause.
  • Broad cleanup unrelated to the demonstrated failure.

Prompt skeleton

Prompt skeleton — deterministic, no model call

Observed behavior
[Include the exact failure and reproduction steps.]

Expected behavior
[Describe the correct result.]

Scope and constraints
[Identify relevant code and regression risks.]

Validation
[Add a regression test and relevant checks.]

From templates/bug.md. Placeholders describe what to fill in; nothing here invents project facts.

Evidence

official

Sources