Wealthsimple data leak is a reminder how important it is to diversify between multiple brokerage 📈

Secureflow v0.0.11 usage looks sticky and moderate🥑. Dropping a new blog post soon 🏄

Today Thinking Machines Lab is launching our research blog, Connectionism. Our first blog post is “Defeating Nondeterminism in LLM Inference” We believe that science is better when shared. Connectionism will cover topics as varied as our research is: from kernel numerics to

New on the Anthropic Engineering blog: writing effective tools for LLM agents. AI agents are only as powerful as the tools we give them. So how do we make those tools more effective? We share our best tips for developers: anthropic.com/engineering/wr…

Secureflow + Qwen3:4b model is blazing fast 🔥 in M2 Max. I'm tired of Claude sonnet rate limits and advising to gradually increase the API usage.

Interesting read! thealliance.ai/blog/building-…

A roundup of what’s new in Claude Code:

+ bike 🚲

Taking in all the fall colours on a ride through Waterloo today. 🍁

Fall officially begins today 🍁

Here's the simplest explanation of Cline's agentic algorithm. It's just a state machine that classifies every request with a tool call into 3 types: 1. Question tools (need clarification) 2. Action tools (gather context) 3. Completion tools (present results) That's it.

I have started adding instructions to CLAUDE.md to stop saying "You're absolutely right" to save some output tokens 🤡

Now you can automate many kinds of browser activity using the new Chrome DevTools MCP. In the other retweets of the base post, many people are already thinking of lots of great use cases for this functionality!

The best visual guide to LLMs you’ll ever see

On monday, I once again demo-ed Secureflow CLI at Builders Club of Waterloo New blog post 👇 codepathfinder.dev/blog/introduci…

Exactly for the same reason and adopted it! 🙌 codepathfinder.dev/blog/introduci…

Experimenting with reinforcement learning (RL) and verifiers is showing great promise for training AI security agents. By guiding models with precise rewards and multi-turn reasoning, we can elevate vulnerability detection beyond pattern matching to adaptive, explainable exploit

Decided to live blog this morning's OpenAI DevDay announcements, since I'm in the audience simonwillison.net/2025/Oct/6/ope…

TIL wandb = "weights and biases" not "wan db" spent two days thinking it was a networking database for storing training logs. 😂

So we will now have 300 pages of a new book called vibe engineering soon 😂 or just rename "coding" to "engineering"