Show Notes
Agent loops let the model pick the next step. Workflows invert that. Code owns the control flow; the model owns the judgment inside each step. Here is the TypeScript file I am running today, type-checked against the live SDK, and the honest answer to whether you should build this now or wait for the official tool. Read the essay: https://vinny.dev/blog/2026-05-26-context-spine-workflow-js/
More from Never the Stack
A change can pass release checks after an agent has exceeded its authority. Platforms need controls during execution, with evidence that those controls hold even when the model follows a malicious instruction. Read the essay: https://vinny.dev/blog/2026-10-12-the-prompt-is-not-the-permission-securing-the-agentic-sdlc/
In May I named the argument underneath a year of writing. Five months later, I'd keep the argument and narrow several claims. Here is what changed, what survived, and what I still can't prove. Read the essay: https://vinny.dev/blog/2026-10-08-the-bottleneck-is-never-the-stack-what-held-and-what-moved/
TypeSafe's Jev gives up prose for typed, probabilistic decisions. The engineering work is defining those decisions, measuring their errors, and deciding when software should ask for help. Read the essay: https://vinny.dev/blog/2026-10-03-jev-is-not-a-faster-llm/
I've used five personal agents to help with the work of everyday life. A restaurant call, a changed appointment, and a playoff-ticket search show what I can hand over, what still needs my decision, and how I know the job is finished. Read the essay: https://vinny.dev/blog/2026-10-01-the-return-of-the-personal-agent/
Building a portable software-development workflow across Claude Code, Codex, Goose, Pi and OpenCode. Read the essay: https://vinny.dev/blog/2026-09-30-the-agent-is-not-the-platform/