SparkBox 1.5.5: AI chat is now opt-out, privacy docs match reality
Posted by tomspark
1.5.5 shipped today with real teeth on privacy controls. AI Troubleshooting now has a one-click toggle in Settings → General that kills the chat widget, blocks /api/chat, and stops anything leaving your box—takes effect instantly, no restart. I've also rewritten what you see in the chat drawer and on privacy.html to be honest about what Anthropic sees (your message + a redacted diagnostic snippet only), what never leaves your server (.env, passwords, license key, module data), and what we store on our side (nothing—the proxy is a passthrough). The policy page now matches what we actually do, not the outdated '12-month retention' copy that was there before. Two fixes round out the release: • Self-update reliability: The dashboard used to show stale JS if the server came back before the container restart. Now both success and restart paths poll /api/auth/status and hard-reload together. • iOS Safari scrolling: min-height:100vh was reflowing the page when the URL bar collapsed mid-scroll. Switched to 100dvh and pushed the topbar onto its own GPU layer so you get smooth scrolling. Head to the Updates tab in your dashboard to pull 1.5.5. Report any issues here in d/sparkbox. --- Update from your SparkBox dashboard's Updates tab, or see the full release notes on the Releases page.