Notes from a CTO who has to put agents into production. Every post says what I ran, what it cost, and whether I'd use it.
I ran Survival LLM's library search with no model on its 22 questions: an on-topic title by the author's pattern for 21, every key fact for only 9.
AI coding agents need more than a manager: coordination, isolation, verification and limited autonomy. Real PRs, real costs, and a 33-line loop to run.
Claude Code has more settings than most of us ever touch. Here are the ones that change a working day, with the catch sitting next to each one.
Mistral Large 4 license read for a five-person startup: can you ship it, can you host the 1,061 GB weights, and is the API cheaper? My maths inside.
A PyTorch upgrade checklist for 2.15: Python 3.10 and old NVIDIA GPUs lose their wheels, and macOS 14 loses MPS. What I checked and what to pin.
MCP skills ship in the official TypeScript SDK 2.4. I ran them, and a hostile server's prompt injection still came back verified: true. Here is why.
Kimi K3.1 Pro: 3T-class, weights promised for Nov 4, not shipped. Price, hardware maths, reactions, and Anthropic SDK code that follows Moonshot's docs.
The Goose programming language has no GC and no lifetimes, and an AI wrote its compiler from a human design. I tried it for an hour. What to trust.
Matt Pocock skills explained: what all 27 plugin skills do, who can call each, how to install, what critics say, and how it compares to GSD and BMAD.
