Model settings
The provider is configured in the backend environment, not in the browser.
Configured provider
Checking…
Analyses still work without a model
Parsing, normalization, grouping and severity ranking are entirely deterministic. Investigations complete with status partial and report those groups without AI hypotheses.
To enable AI analysis, set MODEL_PROVIDER and the matching credentials in the backend's .env, then restart it. See the README for each provider.
Where your logs go
- Logs are processed on your machine by the backend. Nothing is sent to an unconfigured service.
- Redaction runs before evidence is stored or sent, and replaces secrets with placeholders like
[REDACTED_TOKEN]. - Raw logs are not written to the database unless
PERSIST_RAW_LOGSis enabled. - Pattern redaction is best-effort and cannot detect every secret. Review logs before sharing them.