stateUpdate

To persist conversation context across long sessions, use a state update workflow that periodically writes progress, decisions, and learnings into timestamped .state files, ensuring critical information is retained before context resets.

The best way to track decisions and open questions during multi-hour projects is by appending incremental updates to a structured state file, which automatically organizes context, learnings, and next steps into clearly labeled sections.

You create a timestamped state file for conversation progress by running a state update process that enforces deterministic writes to files like .state/{timestamp}.state.md, updating frontmatter and organizing content into structured sections.

Yes, you can append new learnings to an existing state file without losing history by using incremental state updates, which preserve a running history while adding the latest progress, decisions, and context to the active file.

No, capturing evolving context does not require manually formatting sections because the state update workflow automatically organizes entries into predefined sections for context, decisions, learnings, open questions, and next steps.

You should run a state file update periodically during long coaching or project sessions whenever new decisions are made or context evolves, ensuring that essential information is captured before any potential context window reset.