// sessions

Sessions

How Rekord models a recording, where it lives on disk, and how to find your way back to it.

The session model

A session is metadata plus an ordered list of events. Each event carries a relative timestamp, a stream label, and its payload — the same shape that drives replay and every export.

Storage layout

Each recording is a self-contained directory under ~/.rekord/sessions/ (override the root with --root). Nothing is hidden, nothing phones home.

~/.rekord/sessions/<id>/

~/.rekord/sessions/monocron-demo/
├── metadata.json   # name, timing, summary
├── events.jsonl    # append-only event log (output/input/resize)
├── exports/        # generated cast/json/markdown/script/gif/mp4
└── handoff/        # context.md, git.diff, tree.txt, logs.txt

Configuration lives separately at ~/.rekord/rekord.yaml — see Configuration.

Finding sessions

$ rekord list
NAME            COMMANDS  WHEN
api-debug       12        2h ago
monocron-demo   5         yesterday

▋ tip

Replay a session at the timing you recorded it — or speed it up: rekord replay api-debug --speed 2.