🦞🌯 Lobster Roll

Stories by simonw

Watching o3 guess a photo's location is surreal, dystopian and entertaining (simonwillison.net)
2025: The Year in LLMs (simonwillison.net)
Your job is to deliver code you have proven to work (simonwillison.net)
Daft Punk Easter Egg in the BPM Tempo of Harder, Better, Faster, Stronger? (madebywindmill.com)
Something is afoot in the land of Qwen (simonwillison.net)
Vibe code is legacy code (blog.val.town)
Grok: Searching X for "From:Elonmusk (Israel or Palestine or Hamas or Gaza)" (simonwillison.net)
OpenAI are quietly adopting skills, now available in ChatGPT and Codex CLI (simonwillison.net)
My 2.5 year old laptop can write Space Invaders in JavaScript now (GLM-4.5 Air) (simonwillison.net)
Show HN: My LLM CLI tool can run tools now, from Python code or plugins (simonwillison.net)
ChatGPT Containers can now run bash, pip/npm install packages and download files (simonwillison.net)
GPT-5 Thinking in ChatGPT (a.k.a. Research Goblin) is good at search (simonwillison.net)
Related: <i>Google&#x27;s new AI mode is good, actually</i> - <a href="https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=45158586">https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=45158586</a> - Sept 2025 (31 comments)
Useful patterns for building HTML tools (simonwillison.net)
Code like a surgeon (geoffreylitt.com)
Designing agentic loops (simonwillison.net)
AI assisted search-based research works now (simonwillison.net)
Getting DeepSeek-OCR working on an Nvidia Spark via brute force with Claude Code (simonwillison.net)
Let's rename the "vibecoding" tag to "llms"
When Lobste.rs introduced the "vibe coding" tag to mean any post relating to LLMs part of [the justification for that decision](https://lobste.rs/s/lkngrz/new_tag_vibecoding) was this, where the **industry is settling on** link targeted the O'Reilly "Vibe Coding: The Future of Programming" page ([In...
Reverse engineering Codex CLI to get GPT-5-Codex-Mini to draw me a pelican (simonwillison.net)
Your job is to deliver code you have proven to work (simonwillison.net)
New prompt injection papers: Agents rule of two and the attacker moves second (simonwillison.net)
Design Patterns for Securing LLM Agents Against Prompt Injections (simonwillison.net)
Qwen2.5-7B-Instruct-1M and Qwen2.5-14B-Instruct-1M (simonwillison.net)
Stuff we figured out about AI in 2023 (simonwillison.net)
Show HN: Showboat and Rodney, so agents can demo what they've built (simonwillison.net)
Here’s how I use LLMs to help me write code (simonwillison.net)
I can now run a GPT-4 class model on my laptop (simonwillison.net)
Here's how I use LLMs to help me write code (simonwillison.net)
AI should help us produce better code (simonwillison.net)
Whither CockroachDB? (rfd.shared.oxide.computer)