Quick Hits — August 24, 2026

Three smaller stories worth your time: the Python AI-SDK world quietly switching networking layers, a community skill that claws back your own Kindle notes, and the first practical escape hatch drawn for gun-detection software.

Anthropic and OpenAI's Python SDKs are both trading their networking layer today. Simon Willison shipped llm-anthropic 0.27, which he built to be compatible with Anthropic's freshly minted SDK v1.0 — the release that moves the client from the widely used httpx library to its fork, httpx2. OpenAI made the identical switch in its own SDK v3.0 two weeks ago. It matters because nearly every Python app that talks to a frontier model runs on one of these two clients, and both vendors just agreed on the same transport — a quiet signal the ecosystem's plumbing is consolidating. The one-line take: boring dependency news now, groundwork for whatever both labs ship next.

A Claude Code skill recovers the Kindle highlights Amazon's export limit swallows. A developer published a plugin that pulls your own notes out of the Kindle notebook page, then uses the Mac Kindle app's local database and on-device OCR to restore the highlights the export caps truncate or hide entirely. It was proven on four real books — 2,432 highlights extracted, with 815 of them export-blocked and every one recovered. It's macOS-only and must run over Claude's browser-control tooling, so it's a power-user fix rather than a mainstream one. The takeaway: another reminder that "AI agent" skills keep finding genuinely useful data-ownership jobs.

The creator of the first 3D-printed gun says he's built a tool to slip past gun-detection software. Cody Wilson's "Hochulization" scrambles file hashes so hashing-based printer blockers — the approach behind New York's new mandate and a similar California bill — can't recognize a gun file without changing what prints. Wilson claims his team also beat more advanced geometry-analysis models with "adversarial noise," re-saving parts as slightly different meshes and adding unused geometry, and that even a detection model trained on strong data still hit a 30% false-positive rate. It's an early look at the arms race before mandatory print-blocking tech even ships, and a useful stress-test of how durable AI detection really is against a motivated adversary.

Sources: Anthropic SDK migration guide · kindle-highlights skill (GitHub)

Deep Dive — TSMC's record quarter and the repriced order book

TSMC posted the best quarter in its history on Thursday and guided the next one higher, and its American depositary receipts fell anyway. Third-quarter revenue came in at $46.72 billion, up 41.2% year over year and above the top of the company's own guidance range. Net profit rose 68.7% to NT$763.06 billion. The fourth-quarter guide of $52.3 billion to $54.3 billion implies roughly 58% growth at the midpoint against a year earlier. The stock closed at $430.00 the day before the release, down 3.4

AI 101 — What is a chiplet?

A chiplet is a piece of a processor that is manufactured as its own small chip, then packaged alongside other small chips so that the finished part behaves like one big chip. Instead of printing one enormous processor, you print several modest ones and wire them together inside a single package. Every flagship AI accelerator shipping today is built this way. It matters right now because the constraint on AI hardware has moved. TSMC's latest quarter put high-performance computing at 68% of plat

Jev 1.14.0 can read images — if you open its 'Danger Zone'

TypeSafe's decision model learned to look at pictures this morning, but the company shipped the capability behind a warning label instead of a launch announcement. Jev 1.14.0 adds image input, a new question type called Box, and a 262,000-token context window — all three shipped as experimental features behind a program TypeSafe calls the "Danger Zone." The features are off by default and have to be enabled per request through the API; TypeSafe's own documentation says they do not meet product

TSMC's sixth straight record quarter — 58.7% margin, 68% from HPC

The chip cycle's upstream end and its downstream end told two different stories on the same morning: TSMC booked a sixth consecutive record, while the lab that buys its wafers told investors its growth rate is falling fast. TSMC's third-quarter net profit rose 68.7% to NT$763.1 billion (about $23.9 billion), a sixth consecutive record, on revenue of NT$1.494 trillion — up 50.9% year over year and roughly $46.7 billion. Both lines beat the LSEG SmartEstimate consensus (NT$1.46 trillion revenue,

Read more

Deep Dive — TSMC's record quarter and the repriced order book

AI 101 — What is a chiplet?

Jev 1.14.0 can read images — if you open its 'Danger Zone'

TSMC's sixth straight record quarter — 58.7% margin, 68% from HPC