clearcote 0.10.2

Playwright drop-in for the Clearcote anti-fingerprint Chromium build. launch() returns a standard Playwright Browser backed by the verified Clearcote binary, which is auto-downloaded and SHA-256 checked.

clearcote (Python SDK)

A Playwright drop-in for Clearcote — the open, reproducible, anti-fingerprint Chromium build. launch() returns a standard Playwright Browser, so migrating is a one-line import change.

The verified Clearcote binary is auto-downloaded and SHA-256 checked on first use, then cached — no zips or paths to manage.

Platform: Clearcote currently ships a Windows x64 binary, so launch() runs on Windows. (The SDK will download + verify the binary on any OS — handy for packaging — but only launches it on Windows. Linux/macOS builds are on the roadmap.)

Install

pip install clearcote

playwright is pulled in as a dependency. You do not need to run playwright install (Clearcote uses its own browser binary, not Playwright's bundled Chromium).

Usage

from clearcote import launch

browser = launch(
    fingerprint="user-7423",        # per-eTLD+1 seed: same seed => same identity, different => unlinkable
    platform="windows",
    timezone="America/New_York",
    headless=False,
)
page = browser.new_page()
page.goto("https://abrahamjuliot.github.io/creepjs/")
# ... standard Playwright (sync API) from here ...
browser.close()

Already using Playwright? Swap p.chromium.launch(...) for launch(...) from clearcote — the returned object is a normal Playwright Browser. (One shared Playwright driver is started lazily and stopped at interpreter exit.)

Async API (clearcote.async_api)

Inside an asyncio event loop, use the async API — it mirrors the sync one and returns Playwright async objects (the sync API raises "Sync API inside the asyncio loop"):

import asyncio
from clearcote.async_api import launch

async def main():
    browser = await launch(
        fingerprint="user-7423",
        platform="windows",
        timezone="America/New_York",
    )
    page = await browser.new_page()
    await page.goto("https://abrahamjuliot.github.io/creepjs/")
    # ... standard Playwright (async API) from here ...
    await browser.close()

asyncio.run(main())

Same options as the sync launch (fingerprint/persona/proxy/geoip/profile/canvas_bridge/ humanize/…). clearcote.async_api exposes launch, launch_persistent_context, launch_agent, run_agent_task, executable_path, download, plus Profile (use launch(profile="name")). Each launched browser owns its Playwright driver and stops it on await browser.close().

Through a proxy (report the proxy's IP, not your host's)

browser = launch(
    fingerprint="user-7423",
    proxy={"server": "http://host:8080", "username": "u", "password": "p"},  # standard Playwright option
    timezone="America/New_York",
    webrtc_ip="203.0.113.10",       # make WebRTC report the proxy egress IP, not your host's
)

WebRTC won't leak your real IP. The engine fabricates the WebRTC server-reflexive (srflx) candidate at webrtc_ip and sends no real STUN from your host — so WebRTC reports the proxy IP and your real IP never leaks at the packet level. A plain candidate "relabel" doesn't stop the leak (the real STUN packet still goes out from your host); Clearcote sends none. Raw host candidates are suppressed, and the candidate set stays coherent (not empty/disabled).

Auto geo-match (geoip)

Set geoip=True and Clearcote resolves the proxy's exit IP (looked up through the proxy) and auto-fills any unset timezone, accept_language, location, and webrtc_ip so the whole identity — clock, language, geo, and WebRTC IP — matches the proxy's region:

browser = launch(
    fingerprint="user-7423",
    proxy={"server": "http://host:8080", "username": "u", "password": "p"},
    geoip=True,              # timezone, languages, location, AND WebRTC IP all auto-set to the proxy's geo
)

Anything you set explicitly wins over geoip. With no proxy it uses your direct connection's IP. The lookup needs an http(s) proxy — SOCKS proxies are skipped (set timezone/accept_language yourself).

Geo data comes from the offline geoip-all-in-one MaxMind database (downloaded + cached on first use; GPL-3.0 data, the same source Camoufox uses) — more accurate than a single online API — with ip-api.com as a fallback.

Humanized input (humanize, show_cursor)

browser = launch(fingerprint="user-7423", humanize=True)
page = browser.new_page()
page.goto("https://example.com")

page.click("#login")                       # eased bezier glide, then a trusted click
page.fill("#user", "alice")                # focus + key-by-key typing with human timing
page.locator("#pwd").type("s3cr3t")        # locators are humanized too
page.mouse.wheel(0, 800)                   # eased, multi-step scroll
# a held-button drag (slider captchas): the button stays pressed across the move
page.mouse.move(x0, y0); page.mouse.down(); page.mouse.move(x1, y0); page.mouse.up()

humanize=True installs one consistent human-input standard covering moving, clicking, dragging, scrolling and typing — all dispatched as native trusted input (isTrusted === true, navigator.webdriver stays false), at both the page level (page.click/hover/dblclick/ type/fill/press, page.mouse.*, page.keyboard.type) and the locator level (locator.click/type/fill/hover/press_sequentially/drag_to/check/…). Mouse paths are eased, slightly bowed cubic-beziers built from the last cursor position (no snap back to the corner); because they use native input, the button held by mouse.down() stays held across the move, so down → move → up is a real drag (slider captchas work). Clicks get an actionability pre-flight (visible + enabled + stable + not covered) and fall back to the native click if it fails. Typing goes key-by-key with randomized inter-key timing and the occasional correction. page.fill with a value over 200 chars stays atomic (skips per-key typing) to avoid crawling.

show_cursor=True injects a red cursor dot that follows the real mouse, handy for watching a headed run. Both default to off; everything stays standard Playwright when humanize=False.

Persistent profile

from clearcote import launch_persistent_context

context = launch_persistent_context(
    "./profile-7423",
    fingerprint="user-7423",
    platform="windows",
)

Widevine / DRM (widevine=True)

clearcote ships the EME/Widevine plumbing compiled in, but — being 100% open source — it does not bundle Google's proprietary CDM. Pass widevine=True on a persistent context and the SDK fetches that CDM once from Google's own component server (same as a real Chrome receives it), seeds it into the profile, and enables it — so navigator.requestMediaKeySystemAccess('com.widevine.alpha') resolves and DRM streams play, instead of EME being a "no-Widevine" tell.

from clearcote import launch_persistent_context

ctx = launch_persistent_context("./profile-drm", widevine=True)   # fetch + seed + enable the CDM
page = ctx.pages[0] if ctx.pages else ctx.new_page()
page.goto("https://example.com")
ok = page.evaluate("""async () => {
  const a = await navigator.requestMediaKeySystemAccess('com.widevine.alpha',
    [{initDataTypes:['cenc'], videoCapabilities:[{contentType:'video/mp4;codecs="avc1.42E01E"',
      robustness:'SW_SECURE_DECODE'}]}]);
  await a.createMediaKeys(); return true;
}""")
print("Widevine:", ok)            # True
ctx.close()

• Requires a persistent context (the CDM lives in user_data_dir) — not the incognito launch().
• The CDM is cached under ~/.clearcote/WidevineCdm; fetch it ahead of time with fetch_widevine().
• It's opt-in: the clearcote package never distributes Google's CDM — you trigger the download.
• Software-secure (L3) playback. Hardware-secure (L1) paths are out of scope.

AI agent (OpenRouter)

Drive a page with an in-browser AI agent — it perceives the live page, asks an LLM what to do, and executes the steps as real, trusted input through Chrome's Actor framework. Defaults to OpenRouter; switch models with a single slug.

from clearcote import launch_agent, run_agent_task

ctx = launch_agent(
    agent_llm_key=OPENROUTER_API_KEY,        # turns the agent on
    agent_model="openai/gpt-4o-mini",        # any provider/model slug
)
page = ctx.pages[0] if ctx.pages else ctx.new_page()
page.goto("https://example.com")

result = run_agent_task(page, "Click the 'More information...' link.", max_steps=8)
print(result["success"], result["finalText"], result["steps"])
ctx.close()

• agent_llm_key is all you need — the engine auto-enables Chrome's Actor framework (no extra flags).
• agent_llm_url points at any OpenAI-compatible endpoint (default OpenRouter); agent_tool_mode is "tools" (function-calling) or "json".
• Override the model per task: run_agent_task(page, goal, model="anthropic/claude-3.5-sonnet").
• The agent needs a regular profile — use launch_agent / launch_persistent_context, not the incognito launch().

Capture or import a profile

Instead of the synthetic seed-derived identity, you can have Clearcote present a real machine's fingerprint. Pass it to launch() via fingerprint_profile — fields present in the profile override the seed-derived persona; absent fields fall back to the fingerprint seed, so partial profiles stay coherent.

1. Capture from a donor Chrome — open tools/fingerprint-collect/collect.html and click Capture (downloads a JSON), or paste the collector script in DevTools. It records an exhaustive profile (navigator, screen, WebGL, audio, speech voices, fonts, codecs, CSS media, WebGPU, WebRTC). See the collector README.

2. Or convert from the open-source 10k dataset — chrome-fingerprints:

pip install chrome-fingerprints
python tools/fingerprint-collect/convert_dataset.py --out ./profiles --count 100

3. Launch with the profile:

browser = launch(
    fingerprint="seed-1",                 # seeds any field the profile doesn't specify
    fingerprint_profile="profile.json",   # path / dict / JSON string — SDK gzip+base64-encodes it
)

Fingerprint options

All optional. Anything not listed here is passed straight through to Playwright (headless, proxy, args, timeout, slow_mo, …).

Kwarg	Switch	Meaning
fingerprint	--fingerprint	Master seed (per-eTLD+1 farbling root). str or int.
platform	--fingerprint-platform	"windows" | "linux" | "macos".
platform_version	--fingerprint-platform-version	UA-CH platform version.
brand	--fingerprint-brand	"Chrome" | "Edge" | "Opera" | "Vivaldi".
brand_version	--fingerprint-brand-version	Brand version.
gpu_vendor	--fingerprint-gpu-vendor	WebGL UNMASKED vendor.
gpu_renderer	--fingerprint-gpu-renderer	WebGL UNMASKED renderer.
hardware_concurrency	--fingerprint-hardware-concurrency	navigator.hardwareConcurrency.
location	--fingerprint-location	"lat,lng" (only when geo permission is granted).
timezone	--timezone	IANA timezone, e.g. "America/New_York".
accept_language	--accept-lang	navigator.languages + Accept-Language header, e.g. "en-US,en".
webrtc_ip	--webrtc-ip	WebRTC IP to report. The engine fabricates the srflx candidate at this IP and sends no real STUN from the host, so the real IP never leaks (not merely relabeled).
disable_gpu_fingerprint	--disable-gpu-fingerprint	Turn off GPU/WebGL spoofing.
geoip	(directive)	True → resolve the proxy's exit-IP geo and auto-fill timezone/accept_language/location/webrtc_ip.
fingerprint_profile	(directive → --fingerprint-profile)	A real captured machine profile (file path / dict / JSON string); the SDK gzip+base64-encodes it. Fields present override the seed-derived persona; absent fields fall back to fingerprint. Also derives accept_language from the profile's navigator.languages when none is set.
canvas_bridge	(→ --canvas-bridge-*)	Forward canvas/WebGL readbacks to a remote real-GPU host so the pixels a page hashes match the GPU your persona claims. {"url", "auth", "mode", "allow", "deny", "fallback"}; setting url auto-adds --no-sandbox. See docs/CANVAS-BRIDGE.md.
extensions	(→ --load-extension + --disable-extensions-except)	List of unpacked-extension directory paths to load (Chromium forces headed when extensions are present).
humanize	(directive)	True → humanize all input (move/click/drag/scroll/type) as native trusted events, at the page and locator level. See Humanized input.
show_cursor	(directive)	True → inject a red cursor dot that follows the real mouse (handy for watching a headed run).

Headed launches default to no_viewport=True so window.innerWidth tracks the real OS window — an emulated 1280×720 on a real window is an impossible-window tell. Pass an explicit viewport to override.

Proxies: a socks5://user:pass@host:port proxy is routed via --proxy-server (Playwright rejects credentials in its SOCKS descriptor). Chromium can't authenticate SOCKS5, so the credentials are dropped with a warning — put the auth on a local relay.

Saved profiles (Profile)

A Profile bundles a persona (seed, GPU, brand, …) and its canvas_bridge config under one name you can persist and re-launch — the claimed GPU, the bridge endpoint, and the bridge's GPU-keyed cache stay coherent because they travel together.

from clearcote import Profile, launch

# save once
Profile("acct-1", {
    "fingerprint": "acct-1",
    "gpu_vendor": "Google Inc. (Intel)",
    "gpu_renderer": "ANGLE (Intel, Intel(R) UHD Graphics ... D3D11)",
    "canvas_bridge": {"url": "ws://127.0.0.1:9099", "auth": "user:secret"},
}).save()

# re-launch anywhere (explicit kwargs override the saved options)
browser = Profile.load("acct-1").launch(headless=False)
# equivalently: launch(profile="acct-1")

Profiles are JSON at ~/.clearcote/profiles/<name>.json (set CLEARCOTE_PROFILE_DIR to relocate).

API

• launch(**options) → Playwright Browser. Pass profile= (a name, path, or Profile) to launch a saved persona.
• launch_persistent_context(user_data_dir, **options) → Playwright BrowserContext.
• executable_path(executable_path=None, cache_dir=None, quiet=False) → str — resolve (download/verify if needed) the chrome.exe path.
• download(cache_dir=None, quiet=False) → str — pre-fetch + verify without launching.
• Profile — Profile(name, options), .save(path=None), Profile.load(name), .launch(**overrides), .launch_persistent_context(dir, **overrides); plus list_profiles(), load_profile(name).
• RELEASE — the pinned release metadata (tag, version, sha256).

Binary resolution & verification

launch() resolves the browser in this order:

1. executable_path= argument, if given;
2. CLEARCOTE_BINARY environment variable, if set;
3. otherwise download the pinned release, verify its SHA-256 (the hash is baked into this package — it's the trust anchor), extract to a per-version cache, and verify the extracted chrome.exe hash too.

Cache location (override with CLEARCOTE_CACHE):

• Windows: %LOCALAPPDATA%\clearcote\Cache\<tag>
• macOS: ~/Library/Caches/clearcote/<tag>
• Linux: ${XDG_CACHE_HOME:-~/.cache}/clearcote/<tag>

A SHA-256 mismatch is a hard error — the SDK refuses to run an unverified binary. You can independently confirm the published checksums and GPG signatures on the release page.

Stay on the latest build (auto_update)

By default the SDK installs the exact browser build pinned into this package — reproducible, and the baked-in SHA-256 is the trust anchor. To follow new browser releases without upgrading the package every time, opt in:

browser = launch(fingerprint="seed-123", auto_update=True)

or set the environment variable globally:

CLEARCOTE_AUTO_UPDATE=1

With auto_update, the SDK resolves the newest GitHub release, downloads its zip, and verifies it against that release's published SHA256SUMS.txt. When a gpg binary is available it additionally imports the release's public key, confirms its fingerprint equals the pinned CA96F185 F96A693A EDB3AC1F CB00D851 B7A86B0F, and verifies the signed checksum — so an auto-resolved build is cryptographically authenticated, not just downloaded. If GitHub is unreachable it falls back to the pinned release; if the latest release is the pinned one, the audited baked-in hashes are used. Each build is cached per tag, so this only downloads when a new version actually ships. (For locked-down/reproducible deployments, leave auto_update off and bump the package deliberately.)

License

BSD-3-Clause. See LICENSE.