---
title: Concepts
description: The handful of ideas the rest of the platform is built on, and where each one is defined in depth.
sidebar:
  label: The core model
  order: 1
---

Everything in UseAgent reduces to a small set of ideas. Hold these four and the
rest of the system reads cleanly.

1. **The harness lives outside the sandbox**

    The backend is the harness. Runs and steps in Postgres are the source of
    truth. The sandbox runs the engine and the work; the UI renders the log.

2. **The inner loop is rented**

    Engine adapters drive each turn one-shot against a resident engine process
    inside the thread's sandbox and translate each engine's stream into steps.
    No engine web UI, and no engine process outside the sandbox.

3. **Output is canonicalized**

    Every engine's native output is mapped to provider-neutral canonical events,
    so different engines render through one UI contract. Native frames are kept
    for fidelity.

4. **Privileged work goes through the gateway**

    A run never holds long-lived provider or database credentials. It calls
    typed gateway tools that perform privileged work behind short-lived,
    scope-bound capability tokens.

## The pieces

<CardGroup cols={2}>
  <Card title="Runs, threads, and commands" href="/concepts/runs-threads-commands" icon="git-commit-horizontal">
    The units of work: a command starts a run, a run is one turn, a thread is the
    continuing conversation.
  </Card>
  <Card title="Events and streaming" href="/concepts/events-and-streaming" icon="radio-tower">
    Canonical events, native frames, steps, and the two SSE streams that carry
    them to the UI.
  </Card>
  <Card title="Engines and adapters" href="/concepts/engines-and-adapters" icon="plug">
    How Codex, Claude, and OpenCode are driven, the runtime path versus ACP, and
    the Codex subscription relay.
  </Card>
  <Card title="Sandboxes and desktop" href="/concepts/sandboxes-and-desktop" icon="box">
    Isolated Linux workstations, warm pools, the noVNC desktop, and restart
    reconciliation.
  </Card>
  <Card title="Gateway, tools, and approvals" href="/concepts/gateway-tools-and-approvals" icon="shield-check">
    The trust boundary, capability tokens, the in-run tool families, and the
    human approval lane.
  </Card>
  <Card title="Knowledge and learning" href="/concepts/knowledge-and-learning" icon="brain">
    Skills, playbooks, knowledge, wiki, memory, and the human-gated
    self-improvement loop.
  </Card>
  <Card title="Artifacts and workpieces" href="/concepts/artifacts-and-workpieces" icon="file-output">
    Published results, the four editable workpiece kinds, uploads, and the outbox.
  </Card>
</CardGroup>