Nina Labs launches git1file, a tool that packs code repositories for AI workflows
git1file turns a GitHub repository into one compact, structured file for large language models, with at least 25% fewer tokens than existing tools.
Wilmington, Delaware. Nina Labs AI today launched git1file, a tool that turns a GitHub repository into one compact, structured file that a large language model can read in a single pass.
Why we built it
Developers who give a code base to a model run into two limits: the size of the context window and the cost of the tokens. Most of a repository does not help the model. Binaries, logs, lock files and generated code take up space and add nothing.
git1file removes that noise and packs what remains into a format that models parse well. In our tests, it uses at least 25% fewer tokens than existing tools for the same repository, without losing the structure of the code.
Features at launch
- Compression. At least 25% fewer tokens for the same context, so more code fits in the same budget.
- Automatic file exclusion. Binaries, logs and redundant documentation are filtered out.
- XML output. A structured format that makes the file tree and the relationships in the code easy for a model to follow.
- Private repositories. Secure access to private code through the GitHub integration.
- Analysis dashboard. A history of past runs and the token savings for each.
Availability
git1file is available now at git1file.com and needs no setup. To mark the launch, the first 100 supporters could buy a lifetime Pro membership for $10.
The Nina Labs team