Experimenting with intent cells for rerunnable LLM workflows

A developer is experimenting with a syntax called Intent-Cell Coding (ICC DSL) for creating rerunnable LLM workflows, where each step is an "intent cell" that combines readable task text with execution details. The approach aims to address reproducibility challenges by recording exact input versions, hashes, and upstream artifact references, sparking discussion on best practices for maintaining state in LLM demos and workflows.

I am experimenting with a small syntax/workflow idea for multi-step LLM work. The problem I keep running into is that the useful part of an LLM workflow is not just the prompt. It is the trail around it: Chat is good for exploration, but it is a weak record. Scripts are reproducible, but often too rigid for work that starts exploratory. The shape I am testing is a local-first notebook where each step is an “intent cell”. I am calling the syntax ICC DSL, for Intent-Cell Coding. A cell keeps readable task text together with execution details: c1 Collect auto @file -markdown context.md c2 Review fast %from c1 @file -json review.json c3 Final best %from c1 %from c2 @file -markdown final.md The part I am thinking hardest about is file/context identity. A rerun is not really comparable unless the system records exact input versions, hashes, snapshots, retrieval results, or upstream artifact refs. I am curious how people here think about reproducibility for LLM demos and workflows. Do you usually keep this state in notebooks, app logs, model/dataset cards, traces, config files, or generated run manifests?