Docs/Core concepts

Sessions & resume

Sessions are JSONL transcripts on disk with parent-id branching, so you can resume any AI agent conversation at exactly the point it stopped.

Source: crates/gray/src/session_store.rs

Every session is a JSONL file under ~/.gray/sessions. Each line is one message with a parent id, which makes the history a tree rather than a list — branching from an earlier point does not destroy what came after it. An interrupted turn still keeps everything that reached memory.

bash
$gray -c # resume the most recent session
$gray --session <id> # resume one by id (printed when you exit)
$gray resume # pick from a list
$gray resume --last # most recent, no picker
$gray resume --all <id> # by id or prefix, ignoring the cwd filter

Inside the REPL, /resume [id|--last|--all] does the same without leaving. Sessions are filtered by working directory by default, so resuming inside a repo shows that repo's conversations first. A resumed session replays its transcript so you can see where you were.

Housekeeping

bash
$gray sessions prune # delete sessions older than 90 days
$gray sessions prune --older-than-days 30

Join the Discord

Chat with the people building and running Gray. Show what you made with it.

Join Discord →

Need help?

Open an issue on GitHub, run /feedback from the REPL, or ask in Discord.

Open an issue →