Demo mode. Local mock AI.

SECURITY

Security built around your code.

Sensei is designed for a simple promise: code analysis should be useful without making your source code harder to protect. These controls describe the public posture of the product today.

LAST UPDATED - MAY 29, 2026

TRANSPORT

TLS

Encrypted traffic in production

STORAGE

RLS

User-scoped database access in real mode

TRAINING

No

Customer code is not used to train models

CONTROLS

What protects the product

The controls below cover the public app, authenticated surfaces, mock-mode development, and production provider integrations.

Code privacy by default

Submitted code is scoped to the requesting user or workspace. Sensei does not sell submitted code or use customer code to train foundation models.

Encrypted transport

Production traffic is served over TLS. Sensitive application flows are designed around secure browser sessions and protected API boundaries.

Scoped data access

Application data is isolated through repository boundaries, Supabase Row Level Security in real mode, and file-backed scoping in mock mode.

Key separation

Public browser keys, server-only service keys, and AI provider keys are separated through environment configuration.

Input validation

API routes validate request bodies and query parameters before they reach service or repository code.

Account control

Users can export their data, update settings, and request account deletion from the product settings surface.

DEVELOPMENT

Secure development practices

Security is enforced in the implementation path rather than handled as a separate checklist at the end.
  • Validate every public API input before business logic runs.
  • Keep authentication and authorization checks at route boundaries.
  • Avoid logging submitted source code or provider secrets.
  • Maintain separate mock and real provider modes for local development safety.
  • Use dependency, type, lint, and build checks before production changes ship.

CUSTOMERS

How to help keep your account secure

ACCOUNTS

Protect credentials

Use strong passwords, avoid shared accounts, and rotate provider keys if exposure is suspected.

SOURCE CODE

Submit only authorized code

Analyze repositories and snippets you own or are explicitly authorized to inspect.

DISCLOSURE

Report issues quickly

Send security findings through the vulnerability disclosure page.