review-learning-loop

What problem does it solve?

It prevents code review feedback from being wasted as one-off comments by capturing what happened to each comment and converting that outcome into reusable learning.

Core Features & Use Cases

• Outcome collection for every review comment: Tracks accepted, resolved, dismissed, ignored, superseded, and unknown outcomes along with author response and whether code changed.
• False-positive reduction: Identifies dismissed/false-positive patterns by agent and captures candidates that led the agent astray, so future comment conditions can be tightened.
• Reference and team knowledge updating: Creates generalized review references from accepted/resolved judgments while suggesting agent updates using measurable metrics.

Quick Start

Use review-learning-loop after a PR review to submit the PR URL, provider, and a list of comment outcomes so the system can produce learning-loop results with references-to-create, suggested agent updates, and summary metrics.