qirabot 2.4.2

AI vision-driven GUI automation across browsers, mobile, desktop, and games

Qirabot Python SDK

English | 简体中文

Cross-platform GUI automation, driven by multimodal AI vision. Drive browsers, mobile apps, full desktops, and games through pixels — no DOM, no selectors — reaching what frameworks like Playwright, Selenium, and Appium cannot.

Run it standalone (bot.open() launches a browser for you; Android / iOS / Windows-window backends are built in with zero extra dependencies), bolt it onto your existing Playwright / Selenium / Appium / pyautogui session, drop it into a pytest suite, or bind by HWND to drive a Unity / Unreal / native desktop game. Same API across all of them.

📖 Full documentation: qirabot.com/docs (中文)

See it work

https://github.com/user-attachments/assets/649ea80c-63e7-4c85-9ee8-3c8fe17e5ef4

Play an MMORPG from zero to level 15, hands-free — iOS real device. The entire task prompt is one sentence: "This is Fantasy Westward Journey mobile. Create a character, then complete the new-player flow; skip whatever can be skipped." Highlights cut from a single unedited run: full 5:50 video · script

More real, unedited runs — the AI sees only pixels. Click a poster to watch (all demos →):

Clear AFK Journey's tutorial and reach the open world — iOS real device

Play chess on lichess.org — Android real device

Beat a fruit tile-match game on its own — Android real device

Installation

One line — installs uv, qirabot (isolated, never touches your system Python), and Chromium. No pre-installed Python required:

# macOS / Linux
curl -LsSf https://qirabot.com/install | sh

# Windows (PowerShell)
powershell -ExecutionPolicy ByPass -c "irm https://qirabot.com/install.ps1 | iex"

Driving a device instead of a browser? The Android (adb), iOS (WDA), and Windows single-window backends are built into the core package:

uv tool install qirabot        # Android + iOS + Windows window; zero extras

pip, virtualenvs, per-framework extras, and troubleshooting: Installation guide. Whichever path you took, qirabot doctor reports what is installed, what is missing (with the exact fix), and whether your API key reaches the server.

Quick Start

Log in once — this opens your browser to authorize the CLI and saves an API key locally (on a headless server, open the printed URL from any device; --paste enters a key from your dashboard manually):

qirabot login

Then hand the AI a task. Real, unedited output:

$ qirabot browser "Search for SpaceX and get the first sentence of the article" --url wikipedia.org
Task: 6237d4ff-b96b-4c7d-addb-30d8a0334970
[1/20] type_text  ← "SpaceX"
        └ Type 'SpaceX' into the Wikipedia search bar and press enter to search.
Done: Space Exploration Technologies Corp., doing business as SpaceX, is an
      American spaceflight, telecommunications, and artificial intelligence
      company headquartered at the Starbase development site in Starbase, Texas.

Every run writes an HTML report with per-step screenshots; --record captures a video of the whole run.

For sites that need an account, log in once by hand — no API key, no tokens — and every later run that reuses the profile starts already signed in:

qirabot open-browser --user-data-dir ~/.automation --url news.ycombinator.com/login
# log in in the window, close it, then:
qirabot browser "Upvote the top story about Rust" --user-data-dir ~/.automation

Python SDK

The CLI is powered by the same engine. Call bot.ai() from Python and the AI likewise looks at the screen, decides the next action, and loops until the task is done — except the result lands directly in your code, with an on_step callback streaming each action as it happens:

from qirabot import Qirabot, StepResult

bot = Qirabot()
page = bot.open("https://www.wikipedia.org")

def on_step(step: StepResult) -> None:
    label = "done" if step.finished else step.action_type
    print(f"  step {step.step}: {label} {step.params}")

result = bot.ai(page, "Search for SpaceX and get the first sentence of the article", on_step=on_step)
print(f"Success: {result.success}")
print(f"Result: {result.output}")

bot.close()

Prefer to drive each step yourself? The same natural-language targeting works as single-step calls — bot.click(page, "Login button"), bot.extract(...), bot.verify(...) — with your code in control. Need just the coordinates? x, y = bot.locate(page, "the OK button") resolves an element without acting, so you can feed them to your own framework calls.

Bolt onto your existing stack

No rewrite: pass your existing page / driver / device object and mix AI steps with the selectors you already have. Add AI where selectors hurt — visual assertions, dynamic widgets, and flows too tedious to script:

import pytest
from qirabot import Qirabot

@pytest.fixture(scope="session")
def bot():
    with Qirabot(task_name="test-checkout") as bot:   # one task per run
        yield bot

def test_checkout(page, bot):     # `page` is your pytest-playwright fixture
    page.goto("https://shop.example.com")
    page.fill("#username", "test_user")             # your selectors, as-is
    page.click("#login-btn")

    # Visual assertion — survives markup rewrites and CSS refactors
    assert bot.verify(page, "the product grid shows items with prices and no error banner")

    # One line replaces a page of brittle selector steps
    result = bot.ai(page, "Complete checkout, name John Doe zip 10001", max_steps=8)
    assert result.success

Works the same for Selenium, Appium, pyautogui, and the built-in device backends (AdbDevice, WdaClient, Window) — and anything else via a 7-primitive custom adapter.

Domain knowledge: teach the AI your rules

The model knows how to drive a UI — not your game's item names or your team's business terms. Mount reference text for the task and the AI consults it at every step. From the CLI, -k takes a file and repeats, 32KB total:

qirabot browser "Buy 10 stamina potions in the shop" -k game-rules.md -k gm-policy.md

From Python, knowledge takes literal text, a UTF-8 file, or a list mixing both:

result = bot.ai(
    device,
    "Complete every daily quest",
    knowledge=[Path("game-rules.md"), "GM commands may be used once per match"],
)

Knowledge is mounted per call: the next bot.ai() starts clean, so each stage of a long flow carries only what it needs. Two deliberate limits: no URLs — fetch remote sources yourself, so auth and failures stay in your code — and knowledge guides decisions; hard rules like "once per match" belong in custom-tool code (next section), where they can actually be enforced.

Custom tools: let the AI call your code

Mid-task, the AI isn't limited to clicking and typing. custom_tools registers plain Python functions the model can invoke as it works — hit an internal API, query a database, fetch an OTP from your mail server, seed test data, or pause for a human at a CAPTCHA. Name, description, and parameters are introspected from the function itself:

def gm_command(command: str) -> str:
    """Send a command to the game's GM backend and return its reply.
    Available commands: add_energy <amount>, add_gold <amount>"""
    return requests.post(GM_URL, json={"cmd": command}, timeout=10).text

result = bot.ai(
    device,
    "Complete every daily quest. If an out-of-energy popup appears, "
    "use gm_command to add 100 energy and continue",
    custom_tools=[gm_command],
)

The tool runs locally on your machine — the server never sees your endpoints or credentials — and its return value becomes the model's next observation. One instruction now spans systems that used to take a page of glue code: UI steps, backend calls, and human handoffs in a single flow. Details (schemas, error handling, pruning built-in tools): AI Tasks & Custom Tools. Runnable examples: custom_tool_gm.py · 06_human_in_the_loop.py.

Progress overlay

Every CLI task command shows a small always-on-top window in the screen's bottom-right corner: the running instruction, each step's action and reasoning, and the final ✓/✗ outcome. The window is excluded from screen capture (macOS NSWindowSharingNone, Windows WDA_EXCLUDEFROMCAPTURE) and click-through, so it never appears in the bot's own screenshots and never intercepts a click meant for the app below. Turn it off with --no-overlay.

When a task drives the machine's real mouse and keyboard (the desktop backends: Window, pyautogui), a slow-breathing amber glow additionally lines the screen edges for the duration of the run — the "machine is being controlled, hands off" signal, the same visual language as a screen-sharing border. Remote-protocol targets (browser, Android, iOS) don't light it: your mouse stays yours there. The glow is capture-excluded like the window; on Windows versions where exclusion isn't available it simply never shows — glowing bars in every screenshot would blind the bot. --no-overlay turns it off together with the window.

It isn't just bot.ai(): single-step calls (bot.click, bot.press_key, bot.type_text, …) on desktop backends inject real input too, so they light the glow as well — it comes on with the first call and fades a few seconds after the last, so a scripted burst reads as one controlled stretch rather than a flicker.

While the glow is on, a small pill at the top of the screen reads "Hold ESC to stop · 长按 ESC 中止" — hold ESC for about a second to abort the run: the bot stops at the next step boundary (a step may take a few seconds), releases every key and mouse button it was holding, and bot.ai() raises a user_abort error; the task is recorded as cancelled, not failed, so deliberate aborts stay out of your failure metrics. The abort is sticky: every later bot.ai() on the same client raises immediately (no glow, no input), so a try/except around one run can't re-take the machine you just reclaimed — continuing requires an explicit bot.clear_user_abort(). Single-step calls stay available for cleanup. Short ESC taps — yours or the bot's own — never trigger it. The kill switch rides the overlay, so it's off when the overlay is off; on the pyautogui backend, slamming the mouse into a screen corner and leaving it there also aborts (pyautogui's built-in failsafe), overlay or not. On macOS the ESC listener needs the Accessibility permission — the same one desktop control already requires.

In the SDK, one flag covers the common case — the bot runs the window for you: the instruction as the headline with a running-state dot and elapsed clock, each step as step 3/20 · click · "…" plus the model's reasoning, and the final ✓/✗ outcome:

bot = Qirabot(overlay=True)   # every bot.ai() run reports to the window

When your script is more than one bot.ai() call, hold the window yourself: a standalone Overlay displays whatever you tell it, whenever — your own phases included — and ov.step feeds it bot steps for just the AI part:

from qirabot import Overlay

with Overlay() as ov:
    ov.begin("phase 1/3: downloading data…")
    data = download_from_api()                    # your own code, no bot

    ov.begin("phase 2/3: filling in the report system…",
             edge_glow=True)                      # real mouse/keyboard ahead
    bot.ai(pyautogui, "Import the data into the report system",
           on_step=ov.step)                       # bot steps go to the window

    ov.begin("phase 3/3: sending the summary mail…")
    send_email(data)

(Already have an on_step callback of your own? on_step=ov.wrap(my_cb) chains both.)

Platform notes: macOS support installs automatically with qirabot (pyobjc); Windows uses the standard library's tkinter — full capture exclusion needs Windows 10 2004+, older versions show a black box in captures instead of the window content. Everywhere else (Linux, CI, missing GUI) the overlay is a silent no-op: it can never break a run. Runnable example: overlay_progress.py.

Documentation

Topic
Getting started	Installation · Quick Start · CLI Reference
Platforms	Browser · Android (adb, no Appium) · iOS (WDA, no Appium) · Windows & Games (DirectInput) · Desktop · Custom Adapters
Integrations	Playwright · Selenium · Appium · pytest
Advanced	AI Tasks & Custom Tools · Reports & Recording · Configuration · Error Handling
Reference	API — Actions & Platform Matrix

Examples

Runnable examples live in examples/, in three styles:

• Bolt onto your existing tests (pytest) — playwright/, selenium/, appium/, desktop/
• Standalone automation (plain scripts) — scraping / RPA / agents: automation/
• Drive a game — Windows desktop games (bind by HWND) and the iOS MMORPG script behind the demo video: game/

See examples/README.md for which to pick.

Agent Skill

plugins/qirabot/skills/qirabot/ is a pre-built agent skill: an AI agent (Claude Code, Cursor, …) loads it and handles setup, scripting, and verification from a natural-language automation goal. Install in Claude Code:

/plugin marketplace add qirabot/claude-plugins
/plugin install qirabot@qirabot

The skill's reference and templates are drift-tested against the live SDK in CI (tests/test_skill.py). Details: plugins/qirabot/README.md.

Migrating from 1.x (airtest)

2.0 removed the airtest integration; the built-in backends are drop-in replacements (AdbDevice / WdaClient / Window), and a copyable adapter keeps existing airtest scripts running unchanged. Guide: Custom Adapters — Migrating from Airtest. The 1.x series lives on the 1.x branch in maintenance mode — pip install "qirabot<2" always resolves to the newest 1.9.x patch.

License

MIT