nuka.dev

What It Is

Nuky is an open-source developer tool focused on understanding code changes. Its goal is not to replace code review or automatically judge whether code is good or bad, but to help developers see “what changed in the system” when a diff is large—especially when AI generates a large amount of code in one pass. It emphasizes local execution, no cloud dependency, keeping data on the machine, and provides a public Docker image.

Core Capabilities

Functionally, Nuky focuses on structured changes rather than line-by-line text diffs. It shows which parts of a codebase changed the most, whether changes are concentrated or scattered, and outputs stable artifacts such as NUKY_SUMMARY, JSON, and Markdown. Its core signals include CONTRACT_CHANGE, BEHAVIOR_CHANGE, and REFACTOR_ONLY, and it uses mechanisms such as callable identity and dual hashes to distinguish API/contract changes from logical behavior changes. It currently explicitly supports Go and Dart/Flutter, with the Quickstart requiring Go 1.22+ and Dart 3.10+.

Workflow and Integration

The tool is primarily CLI-based. A typical workflow is nuky init, nuky scan, then after modifying code, running nuky diff. CI artifacts are consistently output to .nuky/out/, including scan.json, diff.json, diff.md, and diff_summary.txt. Nuky is explicitly not a policy or enforcement framework; whether a merge should be blocked must be decided by external CI rules or team policies. In terms of ecosystem, there is currently an open-source GitLab repository and a Docker image, while a VS Code extension is still planned.

Pricing, Pros, and Cons

No commercial pricing is mentioned in the main text, only that it is open source, so it can be regarded as free and open-source. Its advantages are that it is privacy-friendly, deterministic, well suited to reviewing AI-generated patches, and able to compress large diffs into readable signals. The limitations are also clear: production language support is only listed for Go and Dart/Flutter; it does not provide quality judgments, so reviewers still need the ability to interpret the output; and information on enterprise collaboration, permissions, support SLAs, and similar areas is missing.

Who It’s For and Access from China

Nuky is suitable for teams that frequently review large patches, use AI coding tools, or want to preserve structured change summaries in CI, especially Go and Flutter projects. There is no information in the main text about access from China, so it needs to be tested in practice. If GitLab access or image pulls are unstable, traditional Git diff, GitHub/GitLab Review, static analysis tools, and self-hosted CI rules can be considered as alternatives or complements.