BNLGit
← Home
Which of your repositories are finished, nearly finished, or dead?
AI is writing more of your code every quarter — and your repository count is exploding with it. Half-built services, abandoned experiments, and agent-generated drafts pile up until nobody can say what is shippable, what is a liability, and what should simply be deleted. BNLGit gives every repo a verdict, with the evidence behind it, on infrastructure you control.
One page. Every repository. A verdict you can defend.
Not a vanity score — a finish-list. Each repo is classed from deterministic signals (reachability, wiring, tests, migrations, churn) before any model opinion is added, and every claim cites its evidence.
The four things buyers actually ask you about.
Nothing leaves your building
Self-hosted on your servers, using your own model keys. No third-party index, no telemetry, no US-cloud data path to explain to a regulator.
The intelligence is a git repo
Summaries are plain, diffable, mirrorable files in git — readable by any tool, and yours forever. They outlive us, and any vendor.
A finish-list, not a dashboard
"Endpoint defined, never registered." "Model written, no migration." Specific, cited gaps a team can close — or delete — this week.
No inference markup
You bring your provider keys and pay them directly. BNLGit records cost and provenance — it never resells you tokens at a margin.
Sovereign by design. BNLGit is built for organisations where GDPR, data-residency, and trade-secret containment are non-negotiable — the code, the summaries, and the model calls all stay inside your perimeter. Built on the open-source Gitea project.