Skip to content

Evidence for review comments

The required reading is the course notes and the studio. These sources offer optional ways to sharpen a disposition; each entry names the primary source and the part to use.

  • Source 1 — Vijayvergiya et al., “AI-Assisted Assessment of Coding Practices in Modern Code Review.” Read the abstract and sections 4.2, 4.3 and 6 in the paper. Use its distinction between correctness and actionability to ask whether a review comment is both true and useful; treat the reported setting as evidence about that study, not a course target. (studies/autocommenter-code-review; as_of: 2026-10-09; Source 1)
  • Source 2 — GitHub staff engineer, “How to review code effectively.” Read “Ask questions” and “To approve or not” in the post. Use the advice to separate a blocking behavior problem from a preference and to request a check for each blocking comment. (studies/github-code-review-philosophy; as_of: 2026-10-09; Source 1)
  • Source 3 — Anthropic, “Writing effective tools for agents.” Read the sections on prototyping and evaluation in the engineering article. Use them to ask what inputs and output evidence a review tool exposes before trusting a result; the article does not establish that any particular product review is correct. (tools/anthropic-effective-tools; as_of: 2026-10-09; Source 1)

For each source, write one sentence that connects its claim to a finding in your studio record and one sentence naming the evidence boundary. Do not report a source's example as your own model run.