See which sessions are running, which need your input and what each one did — across every agent you use. When one gets stuck, hand the session to another. Your sessions never leave your machine.
macOS · Apple Silicon & Intel · Runs locally · No account
Works with 
Features
Every session shows whether it’s running, waiting on you or done. When one needs you, a notification and a badge on the app icon say so, even with the window in the background.
Replies read like documents: Markdown, tables and highlighted code. Reasoning folds to one line, and tool calls stay out of the way until you want them.
Every session you touched today, in one list. Tick one off when it’s done — anything left open waits at the bottom the next morning.
Insights
Sessions, tokens and an estimated cost at API prices, day by day — split by agent, project and model.
Pick a day or a week, and an agent you already use writes up what got done, what’s still open and what’s next.
Handoff
Out of quota or going in circles? Pick another agent. A brief is written from the real transcript, and one command starts the new session where the last one stopped.
→ Local & private
There is no account, no API key and no cloud. Briefs and reviews are written by the agents you already have installed, with whatever they are already signed in to. The app only reads their sessions — its own data lives in ~/.sessionsorted.
Supported agents
Claude CodeCodexCursorCopilotOpenCodePiHermesPricing
Try Pro free for 14 days. No account, no card.
$0
Download $29 $29 one-time
Includes a year of updates. Every version released in that year is yours to keep.
Up to 3 computers · No subscription · Renew updates for $15 a year, or don’t
Everything in Free, plus:
FAQ
No. Briefs and reviews are written by the agents you already have installed, with whatever they are already signed in to.
Claude Code, Codex CLI, Cursor (both the app and its CLI), Copilot (its CLI and desktop app), OpenCode, Pi and Hermes — the ones with a session store on disk. Switch any of them off in Settings if you don’t use it.
Yes. Live status, readable transcripts, search, Today, stats and reviews work the same with one agent as with seven. Handoff is there for the day you add a second.
From the token counts each agent records on your machine, priced at each model’s public API rate. It is what the same work would cost on the API — not what your subscription charges. Cursor doesn’t record token counts, so it has none.
Yes, like any other session: they run on your own agent, under your own account. Nothing runs until you start it.
Once a day, an update check (your platform, version, plan, app language and system language) and a request for current model prices; with Pro, a license check about once a month; feedback only when you send it. Details on the privacy page.
Never. It only reads. Briefs, reviews and its own index live in its own folder, ~/.sessionsorted.
Currently macOS, on Apple Silicon and Intel. Windows and Linux aren’t available. Open Download options to request Windows support and optionally leave your email; there’s no release date yet.
Pro keeps working in every version released before that date — nothing is switched off. Renew for $15 to get another year of new versions, whenever you like. Licenses bought at launch pricing get updates for life.
