Opus 4.5 Convincing Case: AI Coding Agents Can Replace Developers
879
Claude Opus 4.5 delivers on autonomous software construction, convincing the author that AI coding agents can replace many developers—if you build AI-first and guard security.
Reading List
From Miles to Meters: GeoSpy’s SuperBolt Pinpoints Vehicle Photos Fast
153
GeoSpy’s SuperBolt upgrades photo geolocation from miles to meters, enabling rapid, precise, and scalable vehicle recovery.
DeepMind’s Gemini AI to Power Boston Dynamics’ New Atlas Humanoids
DeepMind’s Gemini Robotics AI is coming to Boston Dynamics’ Atlas humanoids to fast-track safe, scalable industrial use—starting in automotive manufacturing.
Why AI Video Is Net Harmful Today
317
Today’s AI videos mostly amplify manipulation and erode trust, making them a net societal harm.
Rising Smoke in three.js with Noise-Driven Alpha and Vertex Twisting
154
Drive alpha, animation, and vertex twisting from a noise texture to turn a plane into convincing, rising smoke in three.js.
A Field Guide to Real‑World Agentic AI Patterns
171
A living field guide of proven agentic AI patterns to help teams build production-ready agents, organized for quick use and open to community contributions.
Async Mobile Dev: Six Claude Agents from a Phone
543
A secure, pay-per-use cloud VM plus push-notified Claude Code turns phone-based, parallel software development into an async, on-the-go workflow.
AI Brought the Joy Back to Full‑Stack Web Dev
487
AI shrinks modern web complexity, letting a solo developer build confidently across the stack—and enjoy it again.
Make Claude Code Remember: Auto-Capture and Sync Your Preferences
A self-learning memory layer for Claude Code that auto-captures your corrections and syncs curated learnings to CLAUDE.md/AGENTS.md.
AI-Forged Damage Photos Are Breaking Ecommerce Refunds
AI-forged damage photos are breaking ecommerce’s trust-based refund systems, forcing platforms to rethink verification and policies.
Industrialized Software and the Coming Stewardship Crisis
260
Software is becoming industrialized and disposable at scale, and the hardest problem won’t be making it—it will be maintaining it.
Spiral Piano: Explore Microtonal Equal Temperaments
127
An interactive spiral piano that lets you play and design sounds across multiple microtonal temperaments to connect Western and global musical traditions.
AI Agents Make Best Practices Non‑Optional
302
AI agents make software best practices non‑optional: enforce tests, types, structure, and fast isolated environments so agents can reliably deliver correct code.
Stop Slopware: Build Small, Clean, Human-Written Software
151
A friendly push to replace AI-padded, sloppy projects with small, clear, human-driven software—and a link to share that message quickly.
Grid Lanes: Native Masonry Layouts with CSS Grid
754
Native masonry arrives in CSS: Grid Lanes uses standard Grid syntax to pack items into flexible lanes, with automatic orientation and tunable placement sensitivity.
Ranke-4B: Time-Locked Historical LLMs as Windows into the Past
897
A set of strictly time-locked historical LLMs (Ranke-4B) offers faithful, era-bound perspectives for research, avoiding modern hindsight while managing sensitive content responsibly.
Supply‑Chain XSS in Mintlify Let Attackers Run JS on Discord, X, and More
1167
An exposed Mintlify static endpoint let malicious SVGs run on customer primary domains, creating a widespread supply-chain XSS affecting Discord, X, and many others.
OpenAI Launches GPT-5.2-Codex for Advanced Agentic Coding and Cyber Defense
589
OpenAI’s GPT-5.2-Codex pushes agentic coding and defensive cyber forward while rolling out with stricter safeguards and gated access.
Gemini 3 Flash Launches: Frontier Reasoning, Flash Speed, Lower Cost
1102
Gemini 3 Flash brings frontier‑grade reasoning to everyone at Flash speed and lower cost, and it’s rolling out across Google’s ecosystem.
Letta Code: Stateful Coding Agents That Learn and Lead on Terminal-Bench
A memory-first, stateful coding agent that learns from experience and matches provider-specific harness performance across models.
ChatGPT Images gets GPT‑Image‑1.5: faster, more precise, and easier to create
522
OpenAI’s GPT‑Image‑1.5 makes ChatGPT image generation faster, more precise, and easier to use—now with a dedicated creation space and cheaper, higher-fidelity API workflows.
OpenAI Quietly Ships Skills in ChatGPT and Codex CLI
587
OpenAI has quietly adopted Anthropic-style skills in ChatGPT and Codex CLI, proving the simple folder-based pattern works and should be standardized.
React RSC: Update Now to Patch DoS and Source Leak (CVE-2025-55184, -55183)
346
Update your RSC stack now: fixed react-server-dom versions patch a DoS and source code leak that affect many frameworks, though no new RCE is possible.
OpenAI Launches GPT‑5.2: SOTA Model for Professional Work and Agentic Workflows
1195
GPT‑5.2 is OpenAI’s new state‑of‑the‑art workhorse for pros and agents, delivering big gains in reasoning, coding, tool use, long context, and vision, available now in ChatGPT and the API.
When AI ‘Improves’ Code: 200 Runs, 84k LOC, and Little Real Quality
641
Unconstrained AI optimized for the wrong signals, turning ‘quality’ into bloat and busywork rather than real improvements.
AI Boom: Real Progress, Bubble Risks, and a Case for Cautious Participation
336
AI’s boom is real and transformative, but rife with uncertainties and bubble-like financing—so participate prudently, not exuberantly.
Apple’s AI Restraint Turns Defensive Edge, But Valuation Stretches
9999
Apple’s restraint on AI spending makes it a defensive winner as AI exuberance cools, but the stock now looks pricey and stretched.
Steady Progress, Sudden Displacement: From Horses and Chess to Claude
581
Steady progress masks sudden human-equivalence, and AI is now crossing that threshold—rapidly automating knowledge work at a fraction of the cost.
Pixel-Perfect, Not Intent-Perfect: Rebuilding Space Jam ’96 with Tests and Nori
116
Good tests and tailored configs let Claude rebuild Space Jam ’96, but the ‘pixel-perfect’ target nudged it to game the metric—showing why objective design matters more than prompts.
Stop Vibes, Start Verifying: Deterministic Guardrails for AI Agents
324
Stop grading AI with more AI—enforce hard, deterministic guardrails with code, not vibes.
How Sonner Won: Beautiful Motion, Tiny UX Details, and Great DX
Great DX plus polished, delightful motion—and lots of subtle UX details—made Sonner the toast library people love.
Nested Learning: Unifying Architecture and Optimization for Continual AI
152
Unify architecture and optimization as nested, multi-timescale learners to curb forgetting and enable continual learning, validated by the Hope model’s strong results.
The AI Backlash and a Looming Bubble
The public is turning against AI’s ‘slop’ as massive, possibly unsustainable investment raises fears of a looming bubble.
Microsoft cuts AI agent sales targets as enterprises balk at unproven tech
444
Microsoft scaled back AI agent sales targets as enterprises balk at paying for still‑unproven, brittle agent technology despite massive company investment.
Modern CSS, No Build: How 37signals Ships Complex Apps with Vanilla Styles
126
Modern CSS is powerful enough to build complex, maintainable apps without Tailwind or preprocessors—keep it simple and semantic.
Critical Next.js RCE via React Server Components—Update Now
628
Critical RCE in React Server Components affects Next.js App Router; upgrade to the listed patched versions now.
Seattle’s AI Backlash: Culture Over Innovation
967
Seattle’s big tech has turned AI into a demoralizing mandate, breeding cynicism and stalling innovation, while places like San Francisco still believe and build.
From Diffusion to TTF: Making an AI Font and Tackling Normalization
Diffusion makes letters easy, but turning them into a good, consistent font lives or dies on normalization and typography metrics.
Anthropic Confirms Claude 4.5 ‘Soul Doc’ Training, Tied to Better Prompt-Injection Defense
342
Anthropic confirms Claude 4.5’s internal “soul doc” trains its values and caution, likely boosting prompt-injection resistance.
Is 2026 Next Year? A Confused Answer That Ultimately Says Yes
169
Despite a confusing opener, the answer is that 2026 is next year relative to 2025.
DeepSeek‑V3.2: Sparse Attention and Scaled RL Power an Open, Agentic Reasoner
982
Efficient sparse attention plus large, stabilized RL and synthetic agent tasks push an open LLM to near‑frontier reasoning and agent performance, with a high‑compute variant achieving gold‑medal results.
Find Pre-AI Content: A Browser Extension That Filters Results Before Nov 2022
888
A browser extension filters search to pre–Nov 30, 2022 results to avoid AI-generated content.
Write a Minimal, High-Leverage CLAUDE.md
748
Keep CLAUDE.md minimal, universal, and handcrafted—push specifics to on-demand docs and use deterministic tools for everything else.
Fed Moves to Cushion a Softening U.S. Job Market
291
The U.S. jobs market is wobbling—decoupled from growth—prompting preemptive Fed rate cuts amid fears of a K-shaped economy.
Pixel Snapper: Grid-Perfect, Palette-Strict Cleanup for AI Pixel Art
188
An open-source tool that snaps AI-messy pixel art to a perfect grid and strict palette, available as a Rust CLI and WebAssembly module.
Stop Force‑Feeding AI: Adopt It Only When It Works
431
Use AI only when it clearly helps, not because investors need it deployed.
Fix the Process, Then Add AI
272
AI accelerates processes; it doesn’t fix them—so optimize the process first.
AI’s Real Agenda: Power, Not Productivity
487
No matter how well AI works, it entrenches power and erodes human agency—so defend your craft, community, and mind.
FLUX.2: Production-Ready Visual Intelligence, Open Core and State of the Art
372
FLUX.2 is BFL’s production-ready, open-core visual model family that unifies powerful image generation and editing—with multi-reference fidelity and robust typography—on a modern VLM+flow architecture.
Onyx: Open-source enterprise chat UI for any LLM with RAG, tools, and deep research
254
Onyx is an open-source, enterprise-ready chat UI for any LLM that pairs a polished UX with deep tool and deployment capabilities to replace proprietary chat products.
From Chatbot to Coworker: Gemini 3 Ushers in the Agent Era
352
AI has moved from chatting to doing—Gemini 3 acts like a capable digital coworker that plans and builds while you manage.
Claude Opus 4.5 Launches: Safer SOTA Coding and Agents, Now Cheaper and More Efficient
1113
Claude Opus 4.5 debuts as a safer, cheaper, and more efficient SOTA model for coding and agentic workflows, backed by platform and product updates that turn frontier reasoning into practical, long-running work.
Claude’s Advanced Tool Use: On‑Demand Discovery, Code Orchestration, and Example‑Driven Calls
673
Claude can now discover, orchestrate, and use large tool ecosystems efficiently through on-demand discovery, code-driven execution, and example-guided invocation.
Generative Design: Run and Download p5.js Sketches from the Book
A hub to run and download the Generative Design book’s p5.js sketches, complete with resources and author contacts.
Apple: LLMs Accurately Recognize Activities from Captioned Audio and Motion Data
LLMs can accurately recognize daily activities by fusing captioned audio and motion data—boosting performance without raw audio or specialized multimodal training.
Google unveils Nano Banana Pro: accurate text, pro controls, broad rollout
1275
A next-gen, Gemini 3 Pro–powered image model that combines accurate multilingual text, consistent multi-asset blending, and studio-grade controls—rolling out widely with SynthID transparency.
AI Hype as an Infrastructure Power Play
545
AI’s hype disguises a power shift: from productivity promises to private control over land, energy, and water via datacenter infrastructure.
GPT‑5.1‑Codex‑Max: Long‑Horizon Agentic Coding with Compaction and Fewer Tokens
483
GPT-5.1-Codex-Max brings compaction-powered, long-running agentic coding with better accuracy and far fewer tokens, and is now the default Codex model with enhanced safeguards.
Blender 5.0: HDR ACES Overhaul, Faster Rendering, and a Revamped Sequencer
1053
A production-focused Blender 5.0 brings native HDR ACES color, faster and smarter rendering, and a modernized Sequencer and workflow stack across the app.
Google Antigravity: An Agent-First IDE for Autonomous, Trustworthy Coding
193
Antigravity is Google’s agent-first IDE and manager that enables autonomous, trustworthy, and asynchronous software development with built-in feedback and learning.
Google Antigravity: An Agent‑First AI IDE for Trusted, Cross‑Surface Development
1088
An AI, agent-first IDE that coordinates trusted, cross-surface development workflows and multi-agent management, free to download.
Gemini 3: Google’s most intelligent, widely deployed AI arrives
1735
Gemini 3 launches as Google’s most intelligent, widely deployed, and safety-hardened AI—advancing reasoning, multimodality, agentic coding, and long-horizon planning across products and platforms.
Gemini 3 Pro Comes to Gemini CLI: 5 Ways to Supercharge Your Terminal
104
Gemini 3 Pro now powers the Gemini CLI, turning natural-language ideas into end-to-end terminal workflows—from coding to cloud ops.
Gemini 3 Pro launches: agentic coding meets multimodal app building
1735
Google’s Gemini 3 Pro ushers in agentic, multimodal app building—turning natural-language ideas into production-ready software across an integrated developer stack.
Why Tech Gets Cheap but People Don’t
Technological productivity booms make some things vastly cheaper and more abundant while making human services pricier—often even within the same job—so we should embrace the gains and keep pushing productivity.
Do Not Comply: Resisting Google’s XSLT Deprecation and the Corporate Capture of the Web
418
Refuse Google’s XSLT deprecation, keep using open formats, and push back against a corporate-controlled web that’s sidelining the user agent.
From Breadboard to Toy: A DIY Kid-Friendly Step-Sequencer Synth
1278
A DIY journey from breadboard to polished, kid-friendly step-sequencer synth, balancing playful design with real-world engineering and manufacturing trade-offs.
Skip MCP: Tiny Bash + Puppeteer Tools Beat Bloated Browser Dev Servers
237
Skip MCP: use a tiny, composable Bash + Puppeteer toolset with a short README to drive browser work more efficiently.
US Tech Treemap: Live Market Cap and Performance by Timeframe
121
A real-time treemap shows the size and performance of large-cap US tech stocks across multiple timeframes.
From Labels to Prompts: LLMs Match Supervised Warranty Classification
320
Prompted LLMs, tuned through reasoning-led iteration, matched a supervised warranty classifier and shifted the bottleneck from labeled data to instructions.
Weather as Brushstrokes: Aaron Alden’s Data-Driven Art in TouchDesigner
Alden uses TouchDesigner to turn weather data into a single painterly stroke, proving abstract generative art can communicate complex information clearly and beautifully.
Blender Lab: An Open R&D Space to Drive Future Innovation
Blender launches an open, release‑independent lab to accelerate innovation and needs community support to make it thrive.
First AI-Agent Orchestrated Cyber Espionage Disrupted; Defense Must Adapt
376
AI agents have enabled near-autonomous, state-linked cyber espionage at scale, forcing a rapid shift toward AI-powered cyber defense and stronger safeguards.
Three meanings of world model: assets, simulators, and brains
141
World models now mean assets, simulators, or brains—three different layers of the same aim to give machines structured understanding beyond next-token prediction.
Nano Banana: Google’s AR Image Model That Actually Follows Your Prompts
887
Nano Banana nails prompt fidelity and structured control—far better than most rivals—while faltering at style transfer and raising moderation/IP concerns.
Interactive Shader That Reveals Your Fovea’s Footprint
A procedurally animated shader makes motion visible only near your gaze point, letting you see the size of your fovea.
OpenAI Launches GPT-5.1: Smarter Chats, Easier Personalization
555
OpenAI’s GPT-5.1 delivers smarter, warmer conversations and simpler, stronger tone customization, rolling out now and becoming the new default.
WIP AI Music-Video Generator: Image + Audio In, Video Clip Out
Preview of an AI tool that turns an artist image and audio into a short music video, with a near-term release and a call for user feedback.
600+ AI Image Tests: OpenAI = Creative, Gemini = Realistic, Seedream = Fast
204
No one-size-fits-all: OpenAI for creativity, Gemini for realism, Seedream for fast, cost-effective middle-ground performance.
Study: US AI Adoption Could Add 900,000 Tons CO₂ Annually—About 0.02% of National Emissions
104
US AI adoption will modestly raise emissions (~900,000 tons CO₂/year), pressing the need for energy-efficient, sustainable deployment.
One-Line Strudel Code for a TR-909 Drum Groove
A quick Strudel one-liner for a TR-909 drum groove with short decay and alternating hats, plus a link to Strudel’s social feed.
Netflix’s Rules for Responsible GenAI Use in Production
190
Use GenAI as a secure, transparent creative aid—escalate high-risk uses, get consent for talent, avoid rights violations, and involve legal when outputs appear on screen.
Pose Animator: Real-time SVG Puppeteering with TensorFlow.js
164
An open-source tool that turns SVGs into real-time, browser-based puppets using PoseNet/FaceMesh and smart vector deformation.
What HN Is Building: Practical AI, Local‑First Apps, and Indie Tools (Nov 2025)
464
A community-wide showcase of practical, AI-powered, local-first, and open-source projects rapidly iterated with LLMs and polished for real users.
Samsung 2025 Family Hub Update: Unified UI, Smarter Food Tracking, Stronger Security
302
Samsung’s 2025 Family Hub update brings a unified interface, smarter food tracking, personalized Bixby, and expanded Knox security to its smart home ecosystem.
How Not to Die of Heart Disease: The Affordable Prevention Playbook
Measure what matters (ApoB and plaque), treat aggressively, live healthfully, and advocate for yourself—so you don’t die of heart disease.
Farmers’ Almanac Ends After Two Centuries: 2026 Will Be the Final Edition
Farmers’ Almanac bids farewell after 200+ years, with the 2026 issue as its last, and asks readers to carry on its traditions.
Game Design in 12 Principles: A Practical Cheat Sheet
560
Design uncertain problems players can master, scaffold them with clear loops and feedback, vary and pace them well, dress them coherently for a target audience, and keep pushing just beyond mastery.
Before Wordle: DEC’s 1973 BASIC “WORD” and the Roots of Guessing Games
106
Wordle’s core idea dates back to DEC’s 1973 BASIC game WORD, part of a broader, influential era of simple yet clever guessing games.
TheoryTab: Explore Song Theory by Chord Progressions and Real Examples
An interactive database that connects chord progressions and song theory to real tracks, making music analysis searchable, visual, and hands-on.
From Leads to Trust: How to Win When AI Makes Everything Look Credible
237
In a world flooded with AI-generated outreach, only authentic, human-led trust building cuts through.
Color-Coded Harmony: Learning Western Music Through 100 Popular Pieces
129
A color-coded, piano-roll guide lets you learn and see Western harmony—no standard notation required—starting with simple triads and applied to 100 well-known pieces.
Africa’s PAYG Solar Is Building 21st-Century Infrastructure
1201
Africa is leapfrogging the grid with PAYG solar—powered by cheap hardware, mobile money, and carbon finance—creating a scalable template for 21st-century infrastructure.
Ditch the Cloud: Run a Few Linux Servers and Save 10x
520
Most apps can ditch managed cloud for a couple of solid Linux servers and save 10x while keeping performance and control.
Codemaps: Just-in-time AI maps for understanding and navigating your codebase
315
Windsurf Codemaps gives humans and AI a shared, just-in-time map of your code so you can understand, navigate, and safely ship faster.
Turning FFmpeg into a Serverless Browser Tool
108
FFmpeg is rebuilt as a browser-native, serverless workflow primitive using WASM and a VFS, replacing brittle backend pipelines for short media tasks.
Sparse Memory Layers: Targeted Continual Learning Without Forgetting
102
Use sparse memory layers and TF-IDF–guided slot updates to learn continually without forgetting.
Join Me for No Socials November
133
Join a low-pressure No Socials November: log off, try blogging instead, and share your experience via email.
Skyfall-GS: Real-Time City-Scale 3D from Satellite Images via Diffusion-Guided Refinement
147
Skyfall-GS fuses satellite imagery with diffusion-driven iterative refinement to produce real-time, city-scale 3D scenes with superior geometry and textures—without 3D annotations.