Just installed

Nice. Your agent has a memory now.

The install mostly runs itself. This page tells you what to expect in the next few minutes and what to try first.

Double-click the installer you downloaded.

No admin password needed. A single progress screen appears and runs for about a minute.

It downloads a small AI model.

Roughly 128MB, used for search on your machine. This takes a few minutes on a normal connection, and progress stays visible the whole time.

Your browser opens your memory dashboard.

When the install finishes, a welcome page opens automatically at localhost:24824/welcome. That dashboard is where you can browse everything Total Recall remembers.

Open your agent and just work.

Open Claude Code or Codex and carry on as usual. Recall starts on its own. Nothing to configure, nothing to save by hand. Using Hermes? Run total-recall hermes setup once and it joins the same memory.

Browser did not open on its own? Open it yourself and go to http://localhost:24824. The dashboard lives there any time you want it.

Your first 5 minutes

Three things to ask your agent.

Copy any of these into Claude Code. Swap the parts in angle brackets for your own topic. These work on day one: the installer indexes your existing session history, so your agent already remembers the work you did before Total Recall.

What did I work on this week? A quick tour of your own recent work, pulled from memory instead of your scrollback.

What did we decide about <topic>? The decision and the reasoning behind it, from the session where it happened.

Remember that <fact> Pins a durable fact. Most things are captured automatically, and this is for the ones you want to make stick.

If something looks off

Quick answers.

The model download is taking a while.

That is normal. The model is about 128MB and the installer shows progress while it downloads. On a slow connection it can take several minutes. Leave it running, it finishes on its own and only happens once.

How do I check it is working?

Ask your agent a memory question, like "what did I work on this week?". If it answers from your past sessions, you are set. You can also open the dashboard at localhost:24824 and watch your sessions appear as you work.

Where does the dashboard live?

In your browser, at http://localhost:24824. It runs on your machine only, so it works offline and nobody else can see it. Bookmark it.

Something else is wrong. Where do I get help?

Open an issue on GitHub and describe what you saw. Include your operating system and roughly what step things went sideways on. We read every report.

Give your agents total recall

98% on LongMemEval, zero tokens at rest, one installer. Your agents remember everything by your next session.