# Quickstart

Install zot and run your first session.

## Install

macOS and Linux:

```shell
curl -fsSL https://www.zot.sh/install.sh | bash
```

Windows (PowerShell):

```powershell
iwr -useb https://www.zot.sh/install.ps1 | iex
```

With Go:

```shell
go install github.com/patriceckhart/zot/cmd/zot@latest
```

## Log in

Use `/login` to store an API key or a subscription credential. zot supports Claude Pro/Max, ChatGPT/Codex, Kimi Code, and GitHub Copilot subscriptions, plus direct API keys for every provider in the catalog.

> **Note on subscription login**
>
> The OAuth client IDs used are the ones published in the official CLIs. Reusing them from a third-party tool may be against their terms of service and may be revoked at any time. The API-key flow is the safe default.

## Run it

```shell
zot # interactive TUI
zot -p "summarize this repo" # one-shot to stdout
cat README.md | zot -p "summarize this text" # combine stdin with the prompt
zot -p --stats stats.json "summarize this repo" # one-shot plus generation stats
zot --json "run the tests" # NDJSON events
```

Pick a model with `/model`, set its reasoning level with `/reasoning`, confine tools for the current session with `/jail`, or enable `jail new sessions by default` in `/settings` to start future sessions jailed. Resume your last session with `zot -c`. Slash command names are case-insensitive, while arguments such as model IDs and paths preserve their original casing.

In print, stream, and JSON modes, piped stdin is prepended to the positional prompt with a newline separator. This lets shell pipelines provide the source text while the positional argument supplies the instruction.

## Print-mode statistics

Use `--stats PATH` with `-p` or `--print` to write generation statistics to a JSON file while keeping stdout limited to the final assistant text. The file is written only after a successful run.

```json title="stats.json"
{
  "provider": "openai",
  "model": "gpt-5",
  "prompt_tokens": 123,
  "reasoning_tokens": 45,
  "generated_output_tokens": 18,
  "elapsed_ms": 1234
}
```

Counts cover every model turn triggered by the prompt, including tool loops. Prompt tokens include cache reads and writes. `reasoning_tokens` is `null` when the provider does not report a separate count; in that case `generated_output_tokens` is the provider's total output count and may include reasoning. Elapsed time covers the agent run, not startup or credential resolution.

## Input navigation

| Key | Action |
| --- | --- |
| `up` / `down` | Move within multi-line input. At the top edge, `up` recalls previous prompts and `down` moves forward through prompt history. |
| `up` / `down` with empty input | Scroll the chat when you are not browsing prompt history. |
| `alt+up` | Move the most recent queued message back into the editor. |
| `ctrl+v` | Paste clipboard text into the focused chat, side chat, dialog, filter, or credential input. On Linux this uses `wl-paste`, `xclip`, or `xsel`; terminal-native paste works without those commands. In the main chat on macOS, image-only clipboard content is attached to the next prompt. |