Repository analysis
From repository to verified deployment.
Zeroth reads the repository, decides whether it can deploy on Zerops, and writes the configuration. Nothing is deployed until you ask for it.
Bring your own LLM key
Used for this run only — held under a short TTL, never stored, discarded when the run settles. BYOK runs skip the shared token budget: your key runs until your provider says stop. Currently supported: OpenAI, Groq, OpenRouter, or any OpenAI-compatible endpoint — more coming.
- check Public GitHub / GitLab
- check Up to 50 MB
- check Isolated deployment
- check Nothing executes during analysis
Nothing has run yet
Analyse a repository above and it will appear here.
| Repository | Status | Action |
|---|
Checking the repository…
Fix these with your own agent
Zeroth does not change your repository. Copy this prompt into Claude, Cursor or whatever you use, apply the changes, then run the analysis again.
"Draft the fix with AI" is the one step where the model reads code — only the files the findings cite, on your request, and the result is a diff for you to apply. Zeroth never writes to your repository.
Nothing has been provisioned.
Configuration ready
Nothing has been deployed yet. Read the files below, then prove they boot.
Used once, then discarded. It is never written to Zeroth's database, its logs, or the downloadable bundle. Failed attempts are torn down; only a project that came up is left standing.
Why these services
Each service traces back to a file in your repository.
Services
Health is only claimed after a verification run.