Skip to main content

​
Code Review Memory

Code Review Memory is a background Template Expert that helps code review adapt to your team’s conventions. When a pull request or merge request is merged, it collects meaningful signals from the review and makes established learnings available to review Experts on future changes. It follows a capture → curate → load workflow:
  1. Capture: It records useful human comments, reactions to agent findings, addressed change requests, and the outcome of the change. Routine acknowledgments, bot updates, and process-only comments are filtered out.
  2. Curate: It combines repeated evidence into concise, repository-specific guidance. Explicit human feedback carries more weight than reactions or an inferred outcome, so strong feedback can become useful memory immediately while weaker signals must recur.
  3. Load: Risk Analyzer, Deep Reviewer, and Pair Reviewer load relevant guidance for the repository and changed paths on their next run. They use it to avoid known false positives and recognize team-specific anti-patterns.
Memory is stored in your organization’s shared VFS and scoped to each repository or project. A raw breadcrumb log preserves the evidence collected from reviews, while a curated knowledge file contains the concise guidance consumed by review Experts. The Memory Manager updates that curated view after each completed change.
Code Review Memory is not the same as creating a skill for every repository. Skills provide explicit, reusable instructions and workflows. Memory is evolving, evidence-backed context that review Experts load automatically for the repository they are reviewing.
  • Best for: Teams that want automated review to learn from recurring feedback and local conventions.
  • How to use: Configure Code Review Memory to run when changes merge, alongside Risk Analyzer or Pair Reviewer. It operates in the background rather than as an interactive Expert.

​
See Also