One episode a day,
the tech that matters today.
1% Tech Digest: AI, open source and industry moves, a new episode every morning.
2026.11
12 episodes
Flutter can finally call native APIs directly, while Claude Code settings files leak secrets on GitHub
Flutter 3.50 and Dart 3.14 ship on the same day, with stable native interop and one frame less touch latency on iOS; Google launches the Gemini Decisions API, which returns probabilities instead of text; Apiiro finds more than a thousand Claude Code settings files with secrets in public repos
Gemini 4 Carbon tops the coding charts while attackers go after the AI keys you store
Google's new model takes first place on CursorBench; Amazon swaps Sonnet for a small model to sort tickets at 98% less cost; Railway and Fireworks were both breached by attackers hunting AI keys
Fable 5.2 gets cheaper while agents collect more and more of our keys
Anthropic's new model matches Opus on coding for about half the cost; ChatGPT Desktop hands your SSH keys to remote agents; a researcher goes from a number plate to unlocking an entire car brand
Apple plugs Xcode into Claude Code while Goldman pulls it from traders
Apple's official plugin lets agents build and test iOS apps; Goldman limits Claude Code to engineers; one heavy user switches to Pi
Your terminal, your chat box, your Mac: all becoming interfaces for agents
Libghostty becomes JetBrains terminal core; Vercel streams generative UI for agents; Apple quietly builds a personal knowledge graph on every Mac
Your login email leaked out, while safer phones got locked out of banking
Claude Code stuffed login emails into User-Agent; Dutch ASN Bank blocks GrapheneOS with Play Integrity
Too-long passwords get you in, and scanners find privilege-escalation paths in bulk
On Alpine, OpenSSH accepts any password longer than 256 characters; OpenAI spent $1.2M in API credits to find 948 Linux privilege-escalation paths
An old PHP bug became remote code execution, and a fake interview emptied an exchange
AI turned a 2024 PHP bug nobody fixed into remote code execution in two weeks, and a fake job interview cost Binance $6.3B
Weight-reading claims get checked, and silent defaults get exposed
Astra's weight-reading claims get taken apart, and people dig into what Postman and Codex changed by default
Models are reading weights, and tools are taking apps apart
Models are reading weights, tools are pulling apps apart, and Anthropic just drew its first line
Verification
AI output keeps growing while the people who verify it don't: 700-plus math papers and a 50-agent civilization experiment
No One Checks the Sender
A shared link takes over ChatGPT accounts; a Linux TCP bug rated CVSS 9.8 is still unpatched in RHEL
