Garnet Guard — Dependency Review Tool

I developed Garnet Guard, a Python tool for reviewing what changed between dependency releases, and built the website that makes its findings understandable. The core idea is to compare an update with the release a project already trusted, then give the developer specific evidence worth reviewing.

The tool’s release-comparison workflow reads package artifacts statically rather than installing or executing them. The garnet diff command surfaces explainable findings, while garnet explain connects a rule to its meaning and a practical review step. Incident replays connect those checks to concrete package changes, with evidence and limitations recorded alongside the output.

My work spans product direction and development priorities, Python implementation, the command-line experience and the public interface used to explore the results. I manage the product as well as building it: defining scope, shaping the developer workflow and keeping its explanations aligned with what the tool does. I paired that work with a JavaScript website, an illustrated identity, a garnet palette and a tabbed replay interface so developers can move from a clear explanation to the evidence behind a finding.

The website brings together replay output, an example pull-request comment, a workflow explanation and a dedicated rule reference. The gallery documents those public views. This is independently owned product work; a replay result is evidence of that specific comparison, not a guarantee that an arbitrary package is safe.

Website and replay examples · Product source and CLI instructions