liveBook · Manning

Comment-driven generation

As explained in Prompt Engineering in Practice

A technique for organizing reviewer feedback as comments anchored to selected content and generating targeted modifications from that feedback.

Comment-driven generation turns reviewer feedback into targeted revision guidance. Anchor each comment to selected content, then batch related comments so they remain in context. Use it to improve specific passages, not to request a full rewrite. section 4.9

A practical workflow:

  • Select the passage that needs work and attach a specific comment explaining the gap. section 4.9
  • Generate a recommendation for each comment, keeping it tied to the original passage and naming what is missing. section 4.9.1
  • Optionally use inversion to generate recommendations, then use refinement to implement them in the draft. section 4.9

Technical RFC review

For a rate-limiting proposal, comments can flag a claim lacking evidence, an unspecified technical approach, and an unclear implementation scope. Turn these into requests for operational metrics, the algorithm and enforcement layer, affected endpoints and thresholds, and rollout, monitoring, and rollback criteria. The RFC is improved when these recommendations become concrete changes. section 4.9.1

Specific comments produce actionable guidance; vague feedback such as “this needs work” tends to produce generic suggestions. The same approach can organize code reviews, security findings, and performance analysis. section 4.9 section 4.9.1

Figure 4.3 Comment-driven generation workflow from anchored feedback to updated draft

Follow the flow from anchored feedback through recommendations to an updated draft. section 4.9

How Prompt Engineering in Practice covers it

How deep it goes
Explains anchoring, batching, and recommendation-to-refinement steps, with inversion as an optional recommendation phase.
Examples
Uses a rate-limiting RFC review to target missing evidence, technical detail, implementation scope, and rollout criteria.
What it assumes
Assumes an existing draft with selected passages and reviewer comments; it does not specify other background requirements.
What you can do after
Practice it by converting comments on selected passages into specific recommendations and implementing them in a draft.
How it connects
Connects with inversion and refinement patterns, and places comment-driven generation among transformational patterns.

Related concepts

This explainer was generated by Manning's knowledge graph from the text of Prompt Engineering in Practice. The section links point to the parts of the book it draws on.