Agentic coding notes from Galapagos Island

It Still Can't Do My Job: Four Years of Moving Goalposts (2022–2026)

The short leash AI coding method for beating Fable

No LLM Code in Dependencies

The gauge broke: devs felt 20% faster with AI, measured 19% slower

Kimi K2.7 Code is generally available in GitHub Copilot

Senior SWE-Bench: open-source benchmark that assesses agents as senior engineers

ZCode: Claude Code from the Makers of GLM

Most rewrites serve the engineer, not the business

Ornith-1.0: self-improving open-source models for agentic coding

Working With AI: A concrete example

The cost YAGNI was never about

GLM 5.2 beats Claude in our benchmarks

Reflections on software engineering in the age of AI

Why current LLM costs are not sustainable

I rewrote PostHog's SQL parser, 70x faster, while barely looking at the code

The Coming Loop

GLM 5.2 vs. Opus

When I reject AI code even if it works

Generative AI Is Having Its Herbalife Moment

DeepSeek V4 Pro at 5% the cost of Claude – what it takes to close the gap

After AI takes everything

SubQ 1.1 Small

SpaceX to buy Cursor AI coding agent operator Anysphere for $60B

Feds freaked over Fable 5 after simple 'fix this code' prompt, not jailbreak

Reviews have become expensive, rewrites have become cheap

AI is code – and can't be prompted into being smarter

Vibe Coder vs. Software Engineer

Ponytail – make your AI agent think like the laziest senior dev in the room

Don't trust large context windows

There is a shadow hanging over this Fable thing

I Am Not a Reverse Centaur

Slightly reducing the sloppiness of AI generated front end

Claude Fable is relentlessly proactive

If you are asking for human attention, demonstrate human effort

MiMo Code is now released and open-source

Grit: Rewriting Git in Rust with Agents

FrontierCode

Launch HN: Intuned (YC S22) – Build and run reliable browser automations as code

DeepSeek V4 Pro beats GPT-5.5 Pro on precision

My automated doubt development process

Office-open-xml-viewer: Office XML document viewer that renders to HTML Canvas

Anthropic/OpenAI may be spending more than $1000 for every $100 you pay them

Tokenomics: Quantifying Where Tokens Are Used in Agentic Software Engineering

Self-hosted dev sandboxes with preview URLs (Docker, Go, no K8s)

Rootshell: A new E2EE email service hosted in Iceland

Agentic Mfw

GitHub Copilot App

Rethinking search as code generation

Chipotlai Max