What problem does Agent Terminal solve?

It bridges voice input, semantic memory, and AI task delegation into one desktop terminal so you can speak a goal and have it routed to the right tool without leaving your workflow.

Quick Start

This is the intended standalone first-run flow.

1. Install Agent Terminal

Download the current platform artifact from the download page, install or launch it, and open Agent Terminal.

2. Open the Composer tab

Type a goal in the Composer input. The pre-send guard runs automatically before dispatch.

3. Check the Runs tab

If Agent Harness is configured, your task appears in the Runs tab with live status.

✦Keep the first request narrow

Ask for one observable outcome, review the proposed action, and inspect the result before starting a second request. Narrow tasks make recovery easier.

4. Browse session history

The History tab shows all past sessions and turns, stored locally in SQLite.

5. Configure Brain and Harness (optional)

Set the environment variables used by the current backend:

BRAIN_BASE_URL=http://localhost:9090
HARNESS_BASE_URL=http://localhost:5001
HARNESS_TOKEN=<your-token>

✦No services required for first run

The terminal works without Brain or Harness on first run. You will see the Composer, History, Context, Transcripts, and Runs tabs — services add enrichment and delegation on top.