Airlock — Privacy Firewall

Airlock is a local-first privacy firewall for text you are about to submit to AI tools, support portals, email replies, forms, and webhooks. Airlock runs only after an explicit action. Paste text into the side panel, read a focused field, scan a selection from the context menu, or start a Pro whole-page scan. It does not continuously monitor browsing or every keystroke. FREE FEATURES • Unlimited manual and focused-field scans • API key, credential, private-key, high-entropy secret, PII, financial identifier, large-paste, and prompt-injection detectors • Local protected-term dictionaries • Per-finding review controls • Redacted safe-output preview PRO FEATURES • Copy or replace safe text • Redaction, pseudonymization, and masked-value transformations • Whole-page and all-form-field scanning • Click a finding to reveal and highlight its source • Apply safe values back to affected form fields • Optional context-sensitive classification using a local Ollama model • Content-free audit events for Splunk, Microsoft Sentinel, Elastic, Datadog, or generic HTTPS • Organization identity stamps and destination-aware routing PRIVACY BY DESIGN Deterministic scanning and safe-text generation happen locally. Ollama support is optional and restricted to a service running on localhost. Airlock does not operate a cloud LLM for scanned content. Webhook reporting is disabled by default. When explicitly enabled, Airlock sends category counts, decision, destination origin, timestamp, and optional organization user/device identifiers to the endpoint selected by the user or administrator. Raw submitted text and matched secret values are not included. Airlock is a guardrail and can produce false positives or false negatives. Review results before submitting sensitive information. Pro is sold by the developer through Lemon Squeezy for $15/month or $199 once. Google is not the seller and does not process the purchase.