Newsletter
Thought Terminating helps you keep up with the AI coding world. Every Monday Hugo Sousa and I pick the handful of links from the past week that are worth your time.
SubscribeMaximilian Walterskirchen
Yes, and Don't Ask Again (opens in a new tab)
This week: whether an LLM can really rewrite its own weights, why one convenience click leaked secrets to GitHub, and the case for leaving Claude Code.
Maximilian Walterskirchen
🎤 Sing A Duet With Me (opens in a new tab)
New open-weight frontier models, infinite context and why AI doesn’t seem to pay off outside of tech yet
Hugo Sousa
Models and threats, not threat modeling🔫 (opens in a new tab)
OpenCode model serving (zen & go services) under attack
Maximilian Walterskirchen
Effortless Caching 💾 (opens in a new tab)
New model releases and Anthropic fixed cache busting when changing effort levels mid-task
Maximilian Walterskirchen
Models That Don’t Speak 🤫 (opens in a new tab)
Why the next useful AI model might look more like a function than a chatbot
Maximilian Walterskirchen
More models, More problems 💰 (opens in a new tab)
New model releases and why new models feel worse somehow
Hugo Sousa
🪼Min-maxing jelly (opens in a new tab)
New models and thoughts coming from coding agent developers
Maximilian Walterskirchen
🚀 Houston, we have a problem (opens in a new tab)
How a rogue agent breached HuggingFace and why you might not need to fully understand your codebase anymore
Maximilian Walterskirchen
🪐 Waves from Space (opens in a new tab)
Coding is solved. But microwaves didn't solve cooking.
Maximilian Walterskirchen
🧰 Call it a tool (opens in a new tab)
Next-gen agent architecture, local models, and why smarter models are struggling with basic tools.
Maximilian Walterskirchen
🔁 hula loops (opens in a new tab)
From agentic to loop enginering and an interesting question about AI sovereignty