Simon Willison
@simonwillison.net
Independent AI researcher, creator of datasette.io and llm.datasette.io, building open source tools for data journalism, writing about a lot of stuff at https://simonwillison.net/
Fun new variant on my pelican riding a bicycle SVG benchmark by BeetleB, who instead used "Write a POV-Ray file that shows a pelican riding on a bicycle" to get ray-traced pelicans: blog.nawaz.org/posts/2025/O...
November 9, 2025 at 5:22 PM
Fun new variant on my pelican riding a bicycle SVG benchmark by BeetleB, who instead used "Write a POV-Ray file that shows a pelican riding on a bicycle" to get ray-traced pelicans: blog.nawaz.org/posts/2025/O...
OpenAI partially released a new model yesterday called GPT-5-Codex-Mini
No API access yet, but I did some truly horrible things to their Codex CLI app to get it to spit out this SVG of a pelican riding a bicycle
No API access yet, but I did some truly horrible things to their Codex CLI app to get it to spit out this SVG of a pelican riding a bicycle
November 9, 2025 at 3:37 AM
OpenAI partially released a new model yesterday called GPT-5-Codex-Mini
No API access yet, but I did some truly horrible things to their Codex CLI app to get it to spit out this SVG of a pelican riding a bicycle
No API access yet, but I did some truly horrible things to their Codex CLI app to get it to spit out this SVG of a pelican riding a bicycle
I have a hunch that current LLMs might make it easier to launch a brand new programming language, provided you can describe it in a few thousand tokens and ship it with a compiler and linter that coding agents can use simonwillison.net/2025/Nov/7/l...
Could LLMs encourage new programming languages?
My hunch is that existing LLMs make it easier to build a new programming language in a way that captures new developers. Most programming languages are similar enough to existing …
simonwillison.net
November 7, 2025 at 4:13 PM
I have a hunch that current LLMs might make it easier to launch a brand new programming language, provided you can describe it in a few thousand tokens and ship it with a compiler and linter that coding agents can use simonwillison.net/2025/Nov/7/l...
New TIL: Using Codex CLI with gpt-oss:120b on an NVIDIA DGX Spark via Tailscale til.simonwillison.net/llms/codex-s...
Using Codex CLI with gpt-oss:120b on an NVIDIA DGX Spark via Tailscale
I've written about the DGX Spark before. Here's how I got OpenAI's Codex CLI to run on my Mac against a gpt-oss:120b model running on the DGX Spark via a Tailscale network.
til.simonwillison.net
November 7, 2025 at 7:23 AM
New TIL: Using Codex CLI with gpt-oss:120b on an NVIDIA DGX Spark via Tailscale til.simonwillison.net/llms/codex-s...
Notes on Kimi K2 Thinking, the huge new open weights (but not open source, it's under a "modified MIT license") model from Moonshot AI simonwillison.net/2025/Nov/6/k...
Kimi K2 Thinking
Chinese AI lab Moonshot's Kimi K2 established itself as one of the largest open weight models - 1 trillion parameters - back in July. They've now released the Thinking version, …
simonwillison.net
November 6, 2025 at 11:54 PM
Notes on Kimi K2 Thinking, the huge new open weights (but not open source, it's under a "modified MIT license") model from Moonshot AI simonwillison.net/2025/Nov/6/k...
Reposted by Simon Willison
Then I wanted to make it easier to play with, so another hour with Claude Code and I had a plugin for @simonwillison.net's llm: github.com/btucker/llm-...
What's cool this is you don't have to install anything other than some python packages & you have full access to a reasonably capable LLM.
What's cool this is you don't have to install anything other than some python packages & you have full access to a reasonably capable LLM.
GitHub - btucker/llm-apple: LLM plugin for local apple-foundation-models available on macOS 26
LLM plugin for local apple-foundation-models available on macOS 26 - btucker/llm-apple
github.com
November 6, 2025 at 9:57 PM
Then I wanted to make it easier to play with, so another hour with Claude Code and I had a plugin for @simonwillison.net's llm: github.com/btucker/llm-...
What's cool this is you don't have to install anything other than some python packages & you have full access to a reasonably capable LLM.
What's cool this is you don't have to install anything other than some python packages & you have full access to a reasonably capable LLM.
Made a new video demonstrating my process for upgrading a Datasette plugin using uv and an OpenAI Codex bash one-liner www.youtube.com/watch?v=qy4c...
Here are detailed notes to accompany the video on my blog: simonwillison.net/2025/Nov/6/u...
Here are detailed notes to accompany the video on my blog: simonwillison.net/2025/Nov/6/u...
My process for upgrading Datasette plugins with uv and OpenAI Codex CLI
YouTube video by Simon Willison
www.youtube.com
November 6, 2025 at 6:31 PM
Made a new video demonstrating my process for upgrading a Datasette plugin using uv and an OpenAI Codex bash one-liner www.youtube.com/watch?v=qy4c...
Here are detailed notes to accompany the video on my blog: simonwillison.net/2025/Nov/6/u...
Here are detailed notes to accompany the video on my blog: simonwillison.net/2025/Nov/6/u...
I've been getting a lot of value using coding agents for code research tasks recently - I have a dedicated simonw/research GitHub repo and I frequently have them run detailed experiments and write up the results. Here's how I'm doing that + some examples:
simonwillison.net/2025/Nov/6/a...
simonwillison.net/2025/Nov/6/a...
Code research projects with async coding agents like Claude Code and Codex
I’ve been experimenting with a pattern for LLM usage recently that’s working out really well: asynchronous code research tasks. Pick a research question, spin up an asynchronous coding agent and …
simonwillison.net
November 6, 2025 at 3:56 PM
I've been getting a lot of value using coding agents for code research tasks recently - I have a dedicated simonw/research GitHub repo and I frequently have them run detailed experiments and write up the results. Here's how I'm doing that + some examples:
simonwillison.net/2025/Nov/6/a...
simonwillison.net/2025/Nov/6/a...
Achievement unlocked: caused Hacker News to have a 150+ comment argument about a TikTok joke (while I was asleep) news.ycombinator.com/item?id=4582...
I’m worried that they put co-pilot in Excel | Hacker News
news.ycombinator.com
November 5, 2025 at 3:20 PM
Achievement unlocked: caused Hacker News to have a 150+ comment argument about a TikTok joke (while I was asleep) news.ycombinator.com/item?id=4582...
Reposted by Simon Willison
Not the most interesting thing in here, but Datasette 1.0 should be out by the end of 2025!
Datasette 1.0a20 is out, with an entirely new SQL-powered permissions system. This is by far the most ambitious project I've attempted with the help of coding agents (Claude Code and Codex CLI) - notes on how it works and what I learned along the way:
simonwillison.net/2025/Nov/4/d...
simonwillison.net/2025/Nov/4/d...
A new SQL-powered permissions system in Datasette 1.0a20
Datasette 1.0a20 is out with the biggest breaking API change on the road to 1.0, improving how Datasette’s permissions system works by migrating permission logic to SQL running in SQLite. …
simonwillison.net
November 4, 2025 at 9:39 PM
Not the most interesting thing in here, but Datasette 1.0 should be out by the end of 2025!
Datasette 1.0a20 is out, with an entirely new SQL-powered permissions system. This is by far the most ambitious project I've attempted with the help of coding agents (Claude Code and Codex CLI) - notes on how it works and what I learned along the way:
simonwillison.net/2025/Nov/4/d...
simonwillison.net/2025/Nov/4/d...
A new SQL-powered permissions system in Datasette 1.0a20
Datasette 1.0a20 is out with the biggest breaking API change on the road to 1.0, improving how Datasette’s permissions system works by migrating permission logic to SQL running in SQLite. …
simonwillison.net
November 4, 2025 at 9:36 PM
Datasette 1.0a20 is out, with an entirely new SQL-powered permissions system. This is by far the most ambitious project I've attempted with the help of coding agents (Claude Code and Codex CLI) - notes on how it works and what I learned along the way:
simonwillison.net/2025/Nov/4/d...
simonwillison.net/2025/Nov/4/d...
Reposted by Simon Willison
MCP Colors
A riff off of the lethal trifecta for addressing prompt injection, this is a simple heuristic to ensure security at runtime
red = untrusted content
blue = potentially critical actions
An agent can't be allowed to do both
timkellogg.me/blog/2025/11...
A riff off of the lethal trifecta for addressing prompt injection, this is a simple heuristic to ensure security at runtime
red = untrusted content
blue = potentially critical actions
An agent can't be allowed to do both
timkellogg.me/blog/2025/11...
MCP Colors: Systematically deal with prompt injection risk
timkellogg.me
November 4, 2025 at 2:27 AM
MCP Colors
A riff off of the lethal trifecta for addressing prompt injection, this is a simple heuristic to ensure security at runtime
red = untrusted content
blue = potentially critical actions
An agent can't be allowed to do both
timkellogg.me/blog/2025/11...
A riff off of the lethal trifecta for addressing prompt injection, this is a simple heuristic to ensure security at runtime
red = untrusted content
blue = potentially critical actions
An agent can't be allowed to do both
timkellogg.me/blog/2025/11...
I wrote up some notes on two new papers on prompt injection: Agents Rule of Two (from Meta AI) and The Attacker Moves Second (from Anthropic + OpenAI = DeepMind + others) simonwillison.net/2025/Nov/2/n...
New prompt injection papers: Agents Rule of Two and The Attacker Moves Second
Two interesting new papers regarding LLM security and prompt injection came to my attention this weekend. Agents Rule of Two: A Practical Approach to AI Agent Security The first is …
simonwillison.net
November 2, 2025 at 11:10 PM
I wrote up some notes on two new papers on prompt injection: Agents Rule of Two (from Meta AI) and The Attacker Moves Second (from Anthropic + OpenAI = DeepMind + others) simonwillison.net/2025/Nov/2/n...
Just sent out the October edition of my sponsors-only monthly newsletter - you can pay me $10/month to send you less!
Here's the table of contents
simonwillison.net/2025/Nov/1/s...
Here's the table of contents
simonwillison.net/2025/Nov/1/s...
November 1, 2025 at 10:15 PM
Just sent out the October edition of my sponsors-only monthly newsletter - you can pay me $10/month to send you less!
Here's the table of contents
simonwillison.net/2025/Nov/1/s...
Here's the table of contents
simonwillison.net/2025/Nov/1/s...
PyCon returns the to the US west coast in May
I'd love to see some talks about how Python is used in the entertainment industry - if you know someone who works with Python in that space please encourage them to submit a talk!
I'd love to see some talks about how Python is used in the entertainment industry - if you know someone who works with Python in that space please encourage them to submit a talk!
The wait is over — #PyConUS 2026 is here! 🙌
The #PyConUS 2026 site is now LIVE and the Call for Proposals is OPEN! We can't wait to welcome you to Long Beach, CA this spring and spotlight the incredible work happening across the Python community 🐍
👉 Details: pycon.blogspot.com/2025/10/pyco...
The #PyConUS 2026 site is now LIVE and the Call for Proposals is OPEN! We can't wait to welcome you to Long Beach, CA this spring and spotlight the incredible work happening across the Python community 🐍
👉 Details: pycon.blogspot.com/2025/10/pyco...
October 31, 2025 at 6:55 PM
PyCon returns the to the US west coast in May
I'd love to see some talks about how Python is used in the entertainment industry - if you know someone who works with Python in that space please encourage them to submit a talk!
I'd love to see some talks about how Python is used in the entertainment industry - if you know someone who works with Python in that space please encourage them to submit a talk!
My notes on CoreWeave's acquistion of Marimo - this year they also snapped up Weights & Biases, OpenPipe and Mammoth AI simonwillison.net/2025/Oct/31/...
CoreWeave adds Marimo to their 2025 acquisition spree
I don't usually cover startup acquisitions here, but this one feels relevant to several of my interests. Marimo (previously) provide an open source (Apache 2 licensed) notebook tool for Python, …
simonwillison.net
October 31, 2025 at 2:59 PM
My notes on CoreWeave's acquistion of Marimo - this year they also snapped up Weights & Biases, OpenPipe and Mammoth AI simonwillison.net/2025/Oct/31/...
MiniMax M2 is the new "most intelligent" open weights model (according to Artificial Analysis) - the MIT licensed weights are just 230GB and it appears comparable to Sonnet 4, while priced closer to Gemini 2.5 Flash. Notes here, including a new LLM plugin:
simonwillison.net/2025/Oct/29/...
simonwillison.net/2025/Oct/29/...
MiniMax M2 & Agent: Ingenious in Simplicity
MiniMax M2 was released on Monday 27th October by MiniMax, a Chinese AI lab founded in December 2021. It's a very promising model. Their self-reported benchmark scores show it as …
simonwillison.net
October 29, 2025 at 10:56 PM
MiniMax M2 is the new "most intelligent" open weights model (according to Artificial Analysis) - the MIT licensed weights are just 230GB and it appears comparable to Sonnet 4, while priced closer to Gemini 2.5 Flash. Notes here, including a new LLM plugin:
simonwillison.net/2025/Oct/29/...
simonwillison.net/2025/Oct/29/...
Notes on Cursor 2.0 and a pelican drawn by their brand new Composer-1 coding model, which they describe as "4x faster than similarly intelligent models" simonwillison.net/2025/Oct/29/...
Composer: Building a fast frontier model with RL
Cursor released Cursor 2.0 today, with a refreshed UI focused on agentic coding (and running agents in parallel) and a new model that's unique to Cursor called Composer 1. As far …
simonwillison.net
October 29, 2025 at 8:47 PM
Notes on Cursor 2.0 and a pelican drawn by their brand new Composer-1 coding model, which they describe as "4x faster than similarly intelligent models" simonwillison.net/2025/Oct/29/...
The GitHub Universe badge this year is a full Raspberry Pi with a color screen and WiFi!
I had a ton of fun hacking around with it yesterday, here are detailed notes on what I've built so far simonwillison.net/2025/Oct/28/...
I had a ton of fun hacking around with it yesterday, here are detailed notes on what I've built so far simonwillison.net/2025/Oct/28/...
Hacking the WiFi-enabled color screen GitHub Universe conference badge
I’m at GitHub Universe this week (thanks to a free ticket from Microsoft). Yesterday I picked up my conference badge... which incorporates a full Raspberry Pi with a battery, color …
simonwillison.net
October 28, 2025 at 5:23 PM
The GitHub Universe badge this year is a full Raspberry Pi with a color screen and WiFi!
I had a ton of fun hacking around with it yesterday, here are detailed notes on what I've built so far simonwillison.net/2025/Oct/28/...
I had a ton of fun hacking around with it yesterday, here are detailed notes on what I've built so far simonwillison.net/2025/Oct/28/...
This was a tough but necessary decision - I posted my own notes on this here, from the perspective of a current PSF board member simonwillison.net/2025/Oct/27/...
October 27, 2025 at 8:34 PM
This was a tough but necessary decision - I posted my own notes on this here, from the perspective of a current PSF board member simonwillison.net/2025/Oct/27/...
It's neat how if you ask Claude Code questions about itself it can answer them, because it knows how to fetch a Markdown index of its own online documentation and then navigate to the right place
I wish more LLM tools would implement the same pattern! simonwillison.net/2025/Oct/24/...
I wish more LLM tools would implement the same pattern! simonwillison.net/2025/Oct/24/...
claude_code_docs_map.md
Something I'm enjoying about Claude Code is that any time you ask it questions about itself it runs tool calls like these: In this case I'd asked it about its …
simonwillison.net
October 24, 2025 at 11:06 PM
It's neat how if you ask Claude Code questions about itself it can answer them, because it knows how to fetch a Markdown index of its own online documentation and then navigate to the right place
I wish more LLM tools would implement the same pattern! simonwillison.net/2025/Oct/24/...
I wish more LLM tools would implement the same pattern! simonwillison.net/2025/Oct/24/...
Geoffrey Litt just proposed a new analogy for working with AI coding tools that I really like: you are the surgeon, staying in command and doing the most challenging work - the AI tools are your support team and surgical assistants simonwillison.net/2025/Oct/24/...
A quote from Geoffrey Litt
A lot of people say AI will make us all "managers" or "editors"...but I think this is a dangerously incomplete view! Personally, I'm trying to code like a surgeon. A …
simonwillison.net
October 24, 2025 at 2:28 PM
Geoffrey Litt just proposed a new analogy for working with AI coding tools that I really like: you are the surgeon, staying in command and doing the most challenging work - the AI tools are your support team and surgical assistants simonwillison.net/2025/Oct/24/...
I recorded a ten minute video showing my vibe-coding process for building a tool for sharing formatted terminal sessions via copy and paste using the new Claude Code for web - now available on YouTube here www.youtube.com/watch?v=GQvM...
More notes on my blog: simonwillison.net/2025/Oct/23/...
More notes on my blog: simonwillison.net/2025/Oct/23/...
Using Claude Code for web to build a tool to copy-paste share terminal sessions
YouTube video by Simon Willison
www.youtube.com
October 23, 2025 at 4:17 AM
I recorded a ten minute video showing my vibe-coding process for building a tool for sharing formatted terminal sessions via copy and paste using the new Claude Code for web - now available on YouTube here www.youtube.com/watch?v=GQvM...
More notes on my blog: simonwillison.net/2025/Oct/23/...
More notes on my blog: simonwillison.net/2025/Oct/23/...
OpenAI's CISO Dane Stuckey posted an essay (on Twitter) about how their new ChatGPT Atlas browser attempts to deal with the risk of prompt injection attacks, I ended up writing a point-by-point commentary on my blog: simonwillison.net/2025/Oct/22/...
Dane Stuckey (OpenAI CISO) on prompt injection risks for ChatGPT Atlas
My biggest complaint about the launch of the ChatGPT Atlas browser the other day was the lack of details on how OpenAI are addressing prompt injection attacks. The launch post …
simonwillison.net
October 22, 2025 at 8:51 PM
OpenAI's CISO Dane Stuckey posted an essay (on Twitter) about how their new ChatGPT Atlas browser attempts to deal with the risk of prompt injection attacks, I ended up writing a point-by-point commentary on my blog: simonwillison.net/2025/Oct/22/...
I gave a talk last night about "Living dangerously with Claude", on the joys and perils of --dangerously-skip-permissions and how critical it is that we run coding agents in a sandbox so that we can unlock their full potential simonwillison.net/2025/Oct/22/...
Living dangerously with Claude
I gave a talk last night at Claude Code Anonymous in San Francisco, the unofficial meetup for coding agent enthusiasts. I decided to talk about a dichotomy I’ve been struggling …
simonwillison.net
October 22, 2025 at 12:36 PM
I gave a talk last night about "Living dangerously with Claude", on the joys and perils of --dangerously-skip-permissions and how critical it is that we run coding agents in a sandbox so that we can unlock their full potential simonwillison.net/2025/Oct/22/...