Session Exporter
Guide
GitHub
Guide
GitHub

Browse & export

Browsing & filtering

Every session from all six tools appears in one list. The left sidebar drives what's shown:

  • Search — matches the title, project, and model.
  • Source — toggle Claude Code / Codex / Cursor / Antigravity / Pi Agent / Kimi Code on and off (each chip shows its session count).
  • Project — the folder a session belongs to. The list is searchable and ordered by session count.
  • Date range — a from/to filter on the session's updated date.

Sort the table by Recently updated, Recently created, Most expensive, Most tokens, Most messages, or Title A–Z.

The top-right counter shows how many of your total sessions match the current filter, along with their combined tokens and cost.

Reading a session

Click a row to open the detail drawer. It shows:

  • The title, source, project path, model, message count, and date range.
  • A token grid: total / input / output, plus cache read, cache write, and reasoning where the source records them, and the cache-hit rate.
  • The estimated cost for that session.
  • The full transcript, with each turn colour-coded by role (user, assistant, reasoning, system) and timestamped. Very long transcripts render the first 500 messages; export to get the complete text.

Selecting

  • Tick a row's checkbox to select it. Selection persists across filters — narrow the list, select more, widen it again, and your earlier picks are still selected.
  • Select all selects every session in the current filter. Untick it (or use Clear) to reset.

When anything is selected, a bar appears at the bottom showing the count, combined tokens, cost, and cache-hit rate, with two export buttons.

Export ZIP

Export ZIP produces a self-contained archive of the selected sessions:

sessions-export.zip ├── index.json # metadata for every session, plus the filter you used ├── index.csv # the same metadata as a flat table ├── README.txt └── sessions/ ├── <source>__<id>.json # full metadata + transcript (machine-readable) └── <source>__<id>.md # the same transcript, human-readable

index.json and index.csv carry all the metadata columns — source, project, dates, model, message count, every token count, cache-hit rate, and cost — so the export is a complete, portable record of what you selected.

Export for Notion

Export for Notion produces a differently-shaped archive designed for Notion's importer — see Export to Notion for the full walkthrough.

Re-scanning

The index is cached for speed. If you've had new sessions since launch, click Refresh in the top bar to re-scan your session files. (Editing prices only? Use ↻ Reload prices in the Stats panel instead — it's instant and doesn't re-scan.)

On this page