Limen settings

Enforcement

Choose how Limen handles sensitive content before it reaches an AI service.

Enable protection Scan before submission Scan pasted text Show in-page scan button Allow bypass in warning mode Keep a local restore map until this tab closes Show deciphered conversations by default. Both views remain private and the AI still receives tags Default mode Block threshold Minimum confidence Maximum scanned characters

Destination policies

One rule per line: hostname-pattern,mode,threshold,bypass. The most specific matching rule wins.

Threshold and bypass are optional. Hard blocks never permit sending the original text.

Organization-sensitive terms

One term per line. Terms remain local unless distributed through managed browser policy.

Detection categories

Enabled categories are scanned. Mandatory-block categories cannot pass unchanged when detected.

Policy simulator

Test the current unsaved policy locally before deploying it.

Destination hostname Sample input

No simulation run.

Audit integrity

Store metadata-only events Retention days

No audit status loaded.

Audit events contain hostname, policy decision, categories, counts, version, and a tamper-evident hash chain. They exclude prompts and detected values.

Configuration portability

Import settings