Trust
What happens to the code you push, in plain terms.
Four things we never do
- Your code is never built.
- Your code is never executed.
- Your code is never used to train models.
- Your code never leaves your tenant.
You decide what we see
The allowlist lives in niiro.yml in your repo. What is not listed never leaves your machine. Environment files, node_modules, .git, lockfiles and binaries over 5 MB are excluded even if a pattern would match them. A secret scan runs before every upload and stops the push on a finding.
Storage
Files are stored encrypted, with a key per project. Blob addresses are derived with a tenant-specific salt, so identical files in different accounts never share a location. Deduplication happens inside one tenant, never across.
Separation
Every row and every job carries its tenant. Analysis and generation run in a sandbox with no code execution and no network access beyond our own APIs. Project tokens are write-only: they can create snapshots but cannot read code or anything derived from it.
Subprocessors
Who processes data on our behalf.
| Service | Purpose | Region |
|---|---|---|
| Vercel | hosting | EU / US |
| Neon | database | EU |
| Railway | workers | EU / US |
| Anthropic | analysis and authoring | US |
| OpenAI | supporting models | US |
| ElevenLabs | voice | US |
| Clerk | accounts | US |
| Stripe | payments | EU / US |
Questions
Security questions go straight to us, not through a form.