Latest writing
Exploring cybersecurity, development, and technology
-
Scion: What Source Control Looks Like at Agent Scale
-
Type-Safe HTML in Zig: How zt Turns Templates Into Checked Code
-
A README Needs a User Test, Not Another Proofread
-
Microsoft Execution Containers Put Agent Permissions Outside the Agent
-
Deno Joins Cloudflare: What Ends, What Continues, and Why celld Matters
-
Whistle Fits Speech Recognition Into 16.9 MB
-
Claude Haiku 5.5: The Small Model Built for Work at Scale
-
JPEG XL in Chrome 155: A New Decoder and a Practical Path to Adoption
-
Gleam 1.19: The Compiler Path from Source to BEAM
-
Cloudflare OHTTP Gateway: Privacy Depends on Who Runs Each Hop
-
Zig 0.17: A Release Guide for Builders and Migrators
-
How Rust Made Its Compiler Faster While Changing Its Core
-
The Request Path Behind Netlify's Faster Edge Functions
-
GPT-6.1 Sol and Dots: A Cheaper Engine and an Agent That Keeps Working
-
Claude Sonnet 5.5: Near-Opus Results at Half the Token Price
-
Cloudflare cf: One CLI for an Entire API
-
Go Concurrency Is a Lifecycle Problem
-
A Local Language Model Can Make Decisions Without Writing Answers
-
Claude Opus 5.5: Frontier Agents Get Faster, Cheaper, and Easier to Trust
-
GPT-6 Sol and Luna Bring Frontier AI Closer to Everyday Work
-
How Python Workers Turn CPython into an Edge Runtime
-
TIN Makes Postgres Search Fast by Speaking Postgres Natively
-
Stop Retry Storms by Making Failures Name Their Owner
-
Inside the One-Month Linux GPU Driver for Apple M4
-
Fugleramme: Turning Birdsong into a Living E-Ink Field Guide
-
Fast Tokio Is a Scheduling Budget, Not a Bag of Tricks
-
Inside the M1 Neural Engine: The Memory Paths Explain the Machine
-
Measuring Code Sloppiness Before It Compounds
-
The Frontier Split: Inside Big Tech's War Over Smarter AI Models and Existential Risk
-
Cracking a 1999 Root Certificate on a Desktop
-
Python in 1024 Bytes: How a Tiny Interpreter Executes Source Directly
-
Cloud in a Bottle: A Personal Cloud That Works Like an App Platform
-
Stop Paying Frontier-Model Prices for Agent I/O
-
Audacity 4: Rebuilding an Audio Editor Without Losing Its Muscle Memory
-
GPT-6 Astra: OpenAI's New Model Is Built to Finish the Work
-
Claude Fable 5.1: Frontier Intelligence Gets Cheaper to Keep Running
-
How RuneScape Turned 56k Limits Into a Multiplayer Protocol
-
How Autocomplete Searches 240 Million Domains Before Keyup
-
What CPython's Official RISC-V Support Actually Guarantees
-
Htmx 4: A Modern Runtime for Server-Driven HTML
-
How Tiny Rust Layout Choices Freed 100 TB of DNS Cache Memory
-
When Unicode Version Drift Becomes a Security Bug
-
Teaching a Language Model to Paint with Editable Code
-
What Changes When Your Executable Is a SQLite Database
-
What an Agent Harness Actually Does
-
Rust Glancer: A Language Server Designed Around a Memory Budget
-
How Real-Time TTS Reaches the First Audible Word in 50 Milliseconds
-
The Silent WebAudio Fingerprint That Hijacked Bluetooth Multipoint
-
Finding One Island Among Millions with Geometry and CUDA
-
Acadia: Bringing Precise Types and Functional Programs to the Database
-
DuckDB 2.0: From Embedded Analytics to a Networked Query Engine
-
How an Experiment Loop Made a GPU Kernel 232x Faster
-
Build Wide, Ship Narrow: Decompose Pull Requests After the Code Teaches You
-
Compression Is Prediction: Why Better Guesses Need Fewer Bits
-
Postgres CDC Without the Connector Tangle
-
Kitesurf: A Browser Built for AI Agents, Not People
-
Why a Smaller Coding-Agent Harness Can Do More
-
Twenty Years of Pandoc: The Architecture Behind a Universal Document Converter
-
Go 1.27: Generic Methods, Better Leak Detection, and a Serious JSON Upgrade
-
Billion-Edge Graph Analytics on a 10 GB Laptop with DataFusion
-
GitHub Stacked Pull Requests: A Practical Review Workflow
-
Demoscene Interfaces: When Creative Tools Had a Point of View
-
Claude Opus 5: A Frontier Model That Is Cheap Enough to Actually Use
-
SIMD for Everyday Programmers: The Parallel Loop Hiding in Plain Sight
-
Parallel Programming Is a Coordination Problem
-
Transcribe.cpp: The Missing Runtime for Local Speech-to-Text
-
Regressive JPEGs: Making One Image Behave Like Video
-
The Human Attention Bottleneck in Agentic Software Development
-
The Real Computers Behind Jurassic Park's UNIX System
-
Build and Ship Apple Apps from the Command Line
-
PgBouncer Scaling Starts When One Core Stops Being Enough
-
GPT-5.6 and GPT-Live: OpenAI Splits Intelligence from the Interface
-
The Bash Script Hidden on a Uniqlo T-Shirt
-
StreetComplete Makes Map Editing Feel Like Field Work, Not Database Work
-
The Log Is the Agent: ActiveGraph and Replayable AI Systems
-
Clean Code Is Now an AI Agent Cost Control
-
Safari MCP Server Makes Browser Agents Cross-Browser by Default
-
Android Developer Verification and the New Gatekeeper Problem
-
Claude Sonnet 5: Anthropic Brings Opus-Like Agentic Power to the Workhorse Tier
-
Age Verification Is Becoming an Identity Layer for Speech
-
You Can Test the Pipeline, Not the Taste
-
F3 Is a Bet That Data Files Should Carry Their Own Decoders
-
Cloudflare Temporary Accounts Make Agent Deployments Disposable
-
Lore: Version Control Built for Game-Scale Repositories
-
The LinkedIn Job Offer That Hid an npm Install Backdoor
-
Software Is Made Between Commits: DeltaDB and the New Shape of Coding Context
-
Claude Fable 5 and Mythos 5: Anthropic's New Split Between General Release and Restricted Power
-
Performative-UI: When Startup Design Tropes Become React Components
-
Why Linear Feels Fast: Local Data, Small Updates, and Product Discipline
-
They're Made Out of Weights: Why AI Feels Stranger Than Software
-
GitHub Token Theft Through a VS Code Webview Bug
-
GitHub Reliability Is Now A Developer Infrastructure Problem
-
Zig's Build System Is Becoming a Two-Process Pipeline
-
Cloudflare's AI Code Review System Is an Orchestrator, Not a Chatbot
-
Claude Opus 4.8: Better Judgment for Long-Running Agentic Work
-
Why I Built MySpec: Stop Prompting and Start Architecting
-
Shamir Secret Sharing: Split Trust Without Splitting Security
-
Jira Is Turing-Complete: When Workflow Automation Becomes a Programming Language
-
AI Chip Costs Are Becoming A Memory Story
-
Structured CSS After Tailwind: What to Keep, What to Leave Behind
-
Writing Code by Hand Is Really About Owning the Architecture
-
Deno 2.8 Makes the Node Compatibility Bet Real
-
Flipper One Is an Open Linux Cyberdeck, Not a Flipper Zero Sequel
-
Agentic Development Lifecycle: Stop Shipping Agents Like Normal Apps
-
GitHub AI Slop Meets The `--author` Loophole
-
AI Will Not Fix a Broken Process
-
AI Is Making Developers Rusty
-
Bambu Lab and the Cost of Burning an Open Source Community
-
TanStack npm Compromise: The Release Pipeline Was the Attack Surface
-
Local AI Needs To Be The Default For App Features
-
LLMs Still Corrupt Documents When You Delegate Real Work
-
SQLite Is a Preservation Format, Not Just an Embedded Database
-
The Bottleneck Was Never Code: An Agent-Era Playbook for Engineering Teams
-
Bun After the Anthropic Acquisition: A Developer Trust Stress Test
-
Your Website Is Not for You: A User-Centered Design Playbook
-
Before GitHub: Open Source Memory, Trust, and What Comes Next
-
Chrome Prompt API: Practical On-Device AI in the Browser
-
Reviving Abandoned Projects with AI Coding Assistants
-
GPT-5.5: OpenAI's Unified Frontier Model for Agents, Code, and Long-Horizon Work
-
Laws of Software Engineering: A Field Guide for Real Teams
-
Claude Design: Anthropic Labs' First Step Into Visual Collaboration
-
Claude Opus 4.7: Anthropic's Most Autonomous Flagship Yet
-
Bring Back Idiomatic Design: Why Consistent Interfaces Still Win
-
MCP vs Skills: Build Connectors, Keep Manuals
-
S3 Files: Why AWS Is Collapsing the File-Object Workflow Gap
-
Quantum Timelines Just Got Shorter: A Practical PQC Migration Playbook
-
The Great Claude Code Leak of 2026: Accident, Incompetence, or the Best PR Stunt in AI History?
-
Pretext: Fast Multiline Text Measurement Without Touching the DOM
-
Inside the LiteLLM PyPI Backdoor: A Minute-by-Minute Incident Response
-
SaaS Is Dead. Long Live SaaaS (Subagent as a Service)
-
GPT-5.4 Mini and Nano: OpenAI's Bet on the Subagent Era
-
How I Write Software with LLMs: A Practical Multi-Agent Workflow
-
When AI Writes Software, Verification Becomes the Real Engineering Work
-
Agents That Run While You Sleep: The Verification Layer Autonomous Coding Needed
-
Go Context Cause: Stop Debugging Blind `context canceled` Errors
-
OpenAI GPT-5.4: The Most Capable Model for Professional Work and Autonomous Agents
-
Clinejection: How One GitHub Issue Title Turned into a Supply Chain Incident
-
Parallel Coding Agents with tmux and Markdown Specs: A Real-World Operating System
-
Decision Trees: Why Nested Rules Still Matter in the LLM Era
-
Microgpt: A ~200-Line Pure Python GPT by Andrej Karpathy
-
Gemini 3.1 Pro: Google's Reasoning Powerhouse Raises the Bar for AI Models
-
Claude Sonnet 4.6: Anthropic's Everyday Workhorse Gets a Major Upgrade
-
GPT-5.3-Codex-Spark: OpenAI's Bet on Real-Time AI Coding Hits 1,000 Tokens Per Second
-
Claude Opus 4.6: Anthropic's New Flagship Pushes the Frontier of Agentic AI
-
Codex App: OpenAI's Command Center for AI Agents on macOS
-
Every Review Layer Multiplies Latency: How Teams Get Slower and How to Recover
-
Claude Cowork: Anthropic's AI Agent That Brings Claude Code Power to Everyone
-
The Ralph Wiggum Technique: A Brief History of Autonomous AI Coding
-
2025: The Year in LLMs - A Comprehensive Review
-
Chrome DevTools MCP (2025): A Practical Guide to AI-Driven Browser Debugging
-
GPT-5.2-Codex: OpenAI's Most Advanced Agentic Coding Model with Cybersecurity Superpowers
-
Gemini 3 Flash: Frontier Intelligence Built for Speed at a Fraction of the Cost
-
GPT-5.2: OpenAI's Breakthrough in Mathematical Reasoning and Coding
-
Claude Opus 4.5: The Most Powerful AI Model for Coding and Agents
-
Dynamic MCPs: Stop Hardcoding Your AI Agent's World
-
Google Gemini 3: Advanced Reasoning and Generative UI Responses
-
GPT-5.1: OpenAI's Smarter, More Conversational AI Model
-
Terminal AI Coding Assistants Compared: Claude Code, GitHub Copilot CLI, and Gemini CLI in 2025
-
Claude Sonnet 4.5: The Best Coding Model in the World
-
GitHub Copilot CLI: AI Coding Directly in Your Terminal
-
GitHub MCP Registry: The Fastest Way to Discover MCP Servers
-
Kiro IDE: Setting a New Paradigm with Spec-Driven Development
-
Vibe Speccing: You're Vibe Coding Wrong, and Here's the Fix
-
Google Gemini CLI: Open-Source AI Agent for Your Terminal
-
Claude Code: The Future of Terminal-Based AI Coding Assistants
-
Cursor IDE: The AI-Powered Code Editor Redefining Developer Productivity in 2025