code-review-graph

Based on the available evidence, code-review-graph is best viewed as promising AI context-reduction infrastructure for technical teams to try, not as a widely proven code review platform yet. Heat signals mainly come from GitHub Trending-style mentions and multiple reposts on X, which show strong attention. Usability evidence is stronger in the official GitHub repo, plus a small number of Zhihu tutorials and demo posts. That means the cautious conclusion is: it looks compelling for large repos, PR review, and AI-assisted code reading, but teams should verify the real gain on their own repositories before treating it as a standard layer.

Its practical role is not to automatically replace human review, nor is it a SAST tool, SonarQube alternative, or classic code quality dashboard. A more accurate analogy is a repo-level retrieval and dependency-pruning layer for AI coding assistants like Claude Code. The evidence says it builds a persistent local code graph with Tree-sitter, then narrows the context to files actually related to a change or question.