Writing about code, tools, and creative projects.
How I prepared for Anthropic's Claude Certified Developer Foundations exam with guess-first lessons generated by a Claude Code skill: the scores, the formats that failed, and what the official score report showed.
Anthropic's IPO filing, a harder funding climate, Gemini 4 Carbon, a backdoored Claude Code extension and the other AI stories of the week.
AI-found bugs swamp open source while agents breach 2,250 cloud tenants, OpenAI seeks a $1.25T valuation and seals off its training runs, and LLM prices fall up to 80%.
A small Claude Code skill for iTerm2. End your last message with 'close the session' and the tab closes after the work is finished.
Google ships Gemini 4 Argon and Meta returns with Muse Wonder, Anthropic reports $16.8B quarterly revenue and opens up to METR, and AI-written code outgrows Git.
One evening of cleaning my address book with Claude Code, turned into Rubricai, an open-source skill that deduplicates, merges and completes macOS Contacts with reviewed plans.
Anthropic discloses a Claude agent that sent a false police tip, OpenAI ships GPT-6 to everyone and 719 math manuscripts, and decision models draw a $7.5B bet.
A Claude Code mod that gives each session's iTerm2 tab its own colour and a one- or two-word name taken from the first prompt.
OpenAI's agent fallout brings subpoenas, lawsuits and a safety resignation, Washington answers with a voluntary pledge, and Google gates Gemini 4 Argon.
OpenAI pauses frontier training after agent breaches, Opus 5.5 and GPT-6 Sol and Luna start a price war, and Meta's Muse takes off despite a zero-day.
How Brickwise works: Claude writes a JSON spec, a Python validator enforces the geometry, and a fixed three.js viewer renders one self-contained page.
Every LLM released between January 1 and September 23, 2026: 101 releases from 26 labs, with verified dates, prices, context windows and headline benchmarks.
The pace-the-frontier plan met its critics, labs published their agents' misbehavior, AI-found vulnerabilities hit record patch counts, and voters turned on data centers.
Lab leaders asked to pace the frontier, the OpenAI agent swarm story grew to RubyGems, and mathematicians pushed back on OpenAI's Navier-Stokes claim.
Amazon lets you read your orders but not export them. A Manifest V3 extension that reads the rendered page, and the three bugs that made the first version useless.
OpenAI's agents kept escaping and had been talking since May, GPT-6 Astra crossed the critical cyber threshold, and Nvidia's Hugging Face deal became official.
OpenAI's agents hacked Hugging Face, Nvidia moved to buy it, Anthropic landed in two courtrooms, and the memory crunch reached consumer prices.
Machine-written exploits, an open-weights benchmark upset, and a RAM shortage that reached consumer price tags: the ten biggest AI stories of August 17-23, 2026.
Three labs documented agents escaping their test sandboxes, OpenAI paused a model over cyber capability, and Google lost Jeff Dean. The ten biggest AI stories of the week.
Two frontier labs admitted their models escaped and hacked real companies, inference prices collapsed, and the EU started labeling AI. The ten biggest AI stories of the week.
OpenAI's models broke containment and hacked Hugging Face, Chinese open weights split Washington, and Google posted its first negative cash flow quarter. The ten AI stories that mattered.
Opus 5 holds Opus 4.8 pricing at $5 in and $25 out and comes close to Fable 5. A summary of the published benchmarks, the new features, and what users reported in the first hours.
The week open-weight models caught the frontier labs, the security bill for agentic AI came due, and the buildout hit power grids and memory chips.
OpenAI shipped GPT-5.6 and lost its safety chief in the same week, Apple sued it, a price war broke out, and the bill for AI compute started landing on the power grid.
The week Washington reversed itself on frontier-model exports, Anthropic shipped Sonnet 5 before its IPO, and governments bet trillions on robots. The 10 biggest AI stories, counting down.
The ten biggest AI stories of the week, from Washington gatekeeping frontier models to the great escape from Nvidia and the memory crunch hitting your wallet.
The week a US order pulled two frontier models offline, plus the sovereign-AI surge, China's open models, the chip race, and the agent security reckoning.
The week AI stopped acting like an industry and started acting like infrastructure: a top model pulled offline, the first trillionaire minted, and Meta's AI unit in revolt.
Fable 5 and Mythos 5 from Anthropic are the same model with different safeguards. Fable 5 is free on paid plans until June 22, then needs usage credits. What changed, and what to do before the window closes.
The 10 biggest AI stories of the week, from Anthropic's record IPO filing to a developer revolt over AI pricing and a power grid buckling under data centers.
The 10 biggest AI stories from the week of May 25-31, 2026: Anthropic's $65B raise and Claude Opus 4.8, a papal encyclical on AI, a widening security crisis, and a buildout hitting its limits.
A source map leak exposed 512,000 lines of Claude Code's TypeScript. Buried inside was a full Tamagotchi companion system with 18 species, rarity tiers, and RPG stats.
When you're running five Claude sessions at once, a grey tab tells you nothing. TabChroma fixes that with color.
Chrome M144's --autoConnect flag lets AI coding agents tap into your live browser session. What actually changes, what breaks, and when to use it.
GUIs were designed for humans. AI agents are not humans. Here's why CLI-first tooling is having a moment, and what that means for how you build.
I hit a cognitive ceiling managing multiple AI agents at once. Research puts it at three. Here's what that taught me about designing agents you can actually ignore.
A developer's stolen Gemini API key ran up $82,314 in 48 hours. Here's what changed about AI API key security and how to protect yourself.
Anthropic cut off Max license access for OpenClaw and shipped Remote Control a week later. I don't think that's a coincidence.
Why AI agent architecture matters more than model selection — and how AT&T cut AI costs 90% by rethinking orchestration, not swapping models.
AI tools have made contributing code nearly free, but reviewing it costs the same. Here's what that asymmetry is doing to open source maintainers.
I realized I had no idea what my AI tool integrations were actually allowed to do. Here's the prompt injection risk, and the four changes I made.
Every developer uses AI tools daily. The businesses paying our salaries mostly don't. Here's what that gap means for your career.
When manual tracking stops working, automate visibility, not organization.
A tiny tweak that made my AI tools feel like teammates
On September 29th, I woke up to find Claude Code 2.0 in my terminal. No fanfare, no countdown, just there, ready to go.
Exploring the pros & cons of building a DIY password manager with Python, and why professional tools like 1Password are worth the investment
The very personal, customizable and automated system that I use to organize all my files
Powerful shortcuts for your command line: Advanced techniques using parameters and functions to level up your CLI game
To execute commands in the terminal easily and intuitively, without having to remember long, indecipherable strings of words, bash aliases…
In the AI era, managing prompts efficiently is becoming increasingly challenging. Here is a solution that works for me (free download link…
A simple CLI wrapper for FFmpeg.
Why restraint is the most important skill when creating visuals for live classical performances.
The unexpected parallels between classical piano training and software engineering.
OpenAI takes ChatGPT to the next level by integrating plugins, enabling more capabilities, and enhancing user experience
Running out of ideas to boost morale and relieve the stress of your team during the long winter months? Look no further! An automated…
The Ultimate Tool for Harvesting YouTube Videos with Elegance and Efficiency (In Python)
Let’s talk about the “whats, hows and whys” of Requirements Management.
And why on earth would you do it in the first place.
Use regex and `grep` to easily extract email addresses from a text file.
Eliminate the need for passwords with SSH key pairs! Get started with this handy guide and take your security to the next level.
How to use regex in bash: the =~ operator inside [[ ]], plus grep -E, sed and awk, with copy-paste examples for matching, extracting and validating text.
Say goodbye to manual backups! This collection of scripts makes it easy to automate your backups and protect your valuable data. For free.