What problem does it solve?
It helps prevent design and implementation mismatches where data ownership, lifetime, and boundary placement become unclear, leading to hard-to-debug correctness issues like stale state, hidden shared mutable state, wrong direction of data flow, and missing teardown.
Core Features & Use Cases
- Boundary and ownership diagnostics: Reviews who creates, owns, consumes, and decides about each non-trivial entity, flagging conceptual scope vs instantiation scope mismatches.
- Topology-first findings: Prioritizes a “draw the data-flow diagram before reading the body” discipline and proposes fixes as reassignment or concrete text-shaped diagrams.
- Diff-aware review workflows: By default reviews against main, or targets a specific file/path, PR number, or git range; can auto-delegate deeper topology questions to a
data-flow subagent.
- Severity-labeled, read-only output: Produces structured findings (with file:line references and diagrams for major/insight) without posting comments.
Quick Start
Ask for a review by saying: run data-flow-review for the current branch against main.