Authenticity of this repository

[gsjewel8-eng]

Does the repo work as it claims, or is it just a bold claim like other repositories which did work but for one response after that take the whole usage away? Does that work the same way, or is this really helpful?

[DeusData]

Thanks for asking this directly.

Short answer: yes, this is the official repository, and yes, the project is meant to be practically useful, not a one response demo.

I also do not want anyone to trust a bold README claim blindly. The right way to evaluate this project is to try it on a real repository you know well and compare its answers with your IDE, grep, or manual code reading.

The tool runs locally. It indexes your code into a local graph and serves that graph through MCP. Your code is not uploaded. The only network check is the normal update check against GitHub releases, so users can see when a newer version exists.

It is useful for structural questions like callers, callees, files, routes, dead code, architecture views, and impact analysis. It is not perfect code understanding. The graph can miss things, especially around language edge cases, generated code, unusual build layouts, and very large projects. We keep those as public issues instead of pretending they do not exist.

If your concern is whether it works once and then becomes useless, that is not the design. The index is persistent, local, and reusable across sessions. Quality depends on the language, project layout, and whether the indexed graph captured the code path you care about.

If you try it and it fails on a real case, please open an issue with the exact query, version, platform, and a small public reproduction if possible. Those reports are useful and we actively triage them.