29 posts · browse by tag
1-bit 1.8 is on the App Store. You can point it at music folders on your device instead of copying them in, watch the music's real waveform behind the progress bar, open available album booklets, and find the records where an artist only guests. Additional improvements include optional offline caching, settings tuned to the DAC you own, and the largest fix pass since 1.0.
Sovereign Shield 0.6.0 is live: an Inspect panel showing your prompt beside what the provider actually gets, plus API-key detection, custom terms and realistic stand-ins. Still entirely local.
Sovereign Shield's browser extension is live on the Chrome Web Store: it redacts Swiss, EU and international identifiers before your prompt reaches ChatGPT, Gemini or Claude.
PaperMC + native Geyser + BedrockConnect. Geyser caught up to my server's Java version, so the ViaVersion translation layer is gone.
My son plays Minecraft on a PlayStation; I wanted him on a Java world I host and control. Here's the whole Bedrock-to-Java bridge — and the DNS trap that eats an afternoon.
A deterministic gateway that lets your team use Gemini or Claude without sending Swiss personal data across the border — tokenised on the way out, restored on the way back.
1-bit 1.7 is on the App Store. Ask Siri to play your library, a Home screen of automatic mixes your own bridge builds from your listening history, performer credits, and a new DAC profile — plus an optional way to pull richer artwork and artist notes. Bridge 0.1.7 ships alongside with Smart Mixes and on-device audio analysis. Same bit-exact DSD over DoP, no transcoding.
I re-ran the same secret-extraction attacks against the flagship models. Gemini 2.5 Pro folds to a forced schema 95% of the time — worse than the cheap tier — while Claude's flagship models drop the prefill surface entirely. The holes move between generations.
Gemini leaks under a forced JSON schema; Claude resists that but folds to prefill. Opposite weak spots, same secret — and one deterministic defense that catches both.
Six hand-written jailbreaks couldn't crack my hardened bot. Then I forced its reply into a JSON schema and it leaked the secret on turn one. Why structured output is the hole.