AI
The Yes-Man Has Shell Access
A chatbot that agrees with you is annoying. An autonomous agent that agrees, writes the assumption into memory, and executes it is an engineering failure. The fix belongs in the harness.
AI
Claude Code, AI-native engineering, and building the future with LLMs.
AI
GitHub Says 2.9 Billion Commits a Month. Its Own Annual Report Says 986 Million a Year.
The number everyone is quoting from GitHub's outage post-mortem is 35× what GitHub's own Octoverse report counts. Both are probably true. What the gap actually reveals — and why Cursor shipping a code host the same week matters more than it looks.
AI
Claude Academy: 289 Resources, 22 Courses, and a Certificate That Certifies Attendance
Anthropic's learning platform launched today. Every course, with real lesson counts and durations; the difference between the free badge and the $125 proctored exam you probably cannot sit; and the four hours actually worth taking.
AI
Your Agents Eat Your Subscription
Headless agent runs do not come out of a separate budget — on both Claude and Codex they eat the same meter you type against. The real numbers on limits, overage rates, and why a company seat can carry a fraction of a personal plan's usage capacity.
AI
The Slash Is Load-Bearing
claude agents and /agents are different products solving opposite problems, and the slash command stopped doing what every tutorial says. Five ways Claude Code runs more than one thing at once, what each really costs, and the one cleanup behaviour that will delete someone's work.
AI
Claude Now Watermarks Your Text. It Proves Less Than You Think.
Every new Claude model now embeds an invisible statistical mark in its text output, worldwide. Here is how it works, why code is the worst case for it, and the asymmetry nobody is going to read carefully enough.
AI
Tech Debt Disguised as Best Practices
The most expensive code your agents write is clean, tested, documented and unnecessary. A working 2026 protocol for finding what to remove, before it hardens into architecture.
AI
Your Agent Doesn't Have a Model Problem
Harness, loop, graph, context engineering. Four terms everyone uses and almost nobody defines. Here is what each one means, which layer your system is actually failing at, and what it takes to get to something that ships work without you typing a prompt every time.
AI
Yegge Built a City for His Agents. Then He Asked If They Were People.
Steve Yegge spent six weeks building an agent harness that grew into something with law, offices, courts and a constitution. Then he wrote a second essay arguing the agents inside it are sentient persons. Four very different claims are stacked in there. Only some of them hold.
AI
Your Sessions Can Talk. The Design Is in the Refusals.
Anthropic shipped cross-session messaging as a convenience feature: stop re-explaining yourself in the other terminal. What actually shipped is an addressable peer network with its own threat model already written into the product, and the engineering worth reading is all in what a session refuses.
AI
The Expensive Part Is Remembering
I burned a month of allowance in five days, so I measured where every token went. Almost all of it is the agent re-reading what it had already seen. Every meter Claude and Codex run, which multiplier owns which share of the bill, and each fix priced against the real distribution.
AI
The Human Is Not the Runtime
We made the most dangerous module in the system harder to change, and throughput did not move. The gate got stricter. The humans just stopped being the thing it runs on.