From 9b4d924e485e568d2f5dbf41f2f682625d3e5c3b Mon Sep 17 00:00:00 2001 From: leeguooooo Date: Thu, 11 Jun 2026 17:34:07 +0900 Subject: [PATCH] docs: make the comparison table honest (CreepJS, Runtime.enable, multi-agent) MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Reader fact-checked the table — three rows overclaimed: - CreepJS: all real-Chrome tools (Claude in Chrome, web-access, us) score ~0%; it is NOT a win vs them. Reframe as "real-browser fingerprint" (✅ for all three, ❌ for Playwright/Puppeteer); note ours is the measured one. - Runtime.enable leak: mark Claude in Chrome "—" (not independently tested) rather than ✅; web-access/Playwright leak, ours is off by default (rebrowser-verified). - Multi-agent: web-access CAN run parallel sub-agents (shared browser), so not ❌. The real differentiator is per-session ISOLATED, command-scoped tab groups. Added footnotes spelling out the caveats. Same fixes in README.zh.md. --- README.md | 10 ++++++---- README.zh.md | 10 ++++++---- 2 files changed, 12 insertions(+), 8 deletions(-) diff --git a/README.md b/README.md index b6ff849..3d11302 100644 --- a/README.md +++ b/README.md @@ -17,12 +17,14 @@ Each alternative solves *part* of the problem. agent-browser-stealth is the only | Works with **any** agent / CLI (not one app) | ❌ Claude only | ✅ | ✅ | ✅ | | Drives your **real, logged-in** Chrome | ✅ | ✅ | ❌ fresh empty profile | ✅ | | **No "Allow remote debugging?" popup** | ✅ | ❌ every connect | — (own browser) | ✅ native messaging | -| **Undetectable** (CreepJS) | ~real | ~real | ❌ automation markers / headless | ✅ **0% stealth, verified** | -| `Runtime.enable` CDP leak (rebrowser) | — | leaks | leaks | ✅ **off by default** | -| **Concurrent agents**, isolated tab groups | ❌ | ❌ | ❌ | ✅ | +| Real-browser fingerprint (CreepJS ~0%)¹ | ✅ | ✅ | ❌ automation markers / headless | ✅ **verified 0%** | +| **No `Runtime.enable` CDP leak** (rebrowser)² | — | ❌ leaks | ❌ leaks | ✅ **off by default** | +| Many agents on **one** real Chrome, isolated tab groups³ | ❌ single app | ⚠️ shared tabs, no isolation | ❌ separate browsers | ✅ | | Permissions footprint | 16 incl. `` | full CDP | full control | **7, no ``** | -> Short version: **Claude in Chrome** is great but locked to Claude; **web-access / raw `--remote-debugging-port`** pops Chrome 136+'s consent dialog on every attach; **Playwright/Puppeteer/browser-use** launch a *fresh* browser (no login, automation-flagged, often headless → detected). agent-browser-stealth drives the Chrome you're already signed into, over a native-messaging extension, and reads as a 100% real human browser. Evidence in [Anti-detection](#anti-detection). +> Short version: **Claude in Chrome** is great but locked to Claude; **web-access / raw `--remote-debugging-port`** pops Chrome 136+'s consent dialog on every attach; **Playwright/Puppeteer/browser-use** launch a *fresh* browser (no login, automation-flagged, often headless → detected). agent-browser-stealth drives the Chrome you're already signed into, over a native-messaging extension, and reads as a 100% real human browser. +> +> ¹ All three real-Chrome tools score ~0% on CreepJS (it's a real browser); we've measured ours. ² rebrowser's `runtimeEnableLeak` — verified clean on our relay path; Claude in Chrome not independently tested (—). ³ web-access can run parallel sub-agents on one browser, but without per-session isolation; each `--session` here gets its own colored, command-isolated tab group. See [Anti-detection](#anti-detection) for the measured numbers. ## Why this fork? diff --git a/README.zh.md b/README.zh.md index 3b95891..1579818 100644 --- a/README.zh.md +++ b/README.zh.md @@ -17,12 +17,14 @@ | **任意** agent / CLI 都能用（不绑单一 app） | ❌ 仅 Claude | ✅ | ✅ | ✅ | | 驱动你**真实、已登录**的 Chrome | ✅ | ✅ | ❌ 全新空 profile | ✅ | | **不弹 "Allow remote debugging?"** | ✅ | ❌ 每次连都弹 | —（自带浏览器） | ✅ 原生消息 | -| **不可检测**（CreepJS） | ~真实 | ~真实 | ❌ 自动化特征 / headless | ✅ **0% stealth，实测** | -| `Runtime.enable` CDP 泄漏（rebrowser） | — | 泄漏 | 泄漏 | ✅ **默认关闭** | -| **多 agent 并发**、标签组隔离 | ❌ | ❌ | ❌ | ✅ | +| 真实浏览器指纹（CreepJS ~0%）¹ | ✅ | ✅ | ❌ 自动化特征 / headless | ✅ **已实测 0%** | +| **无 `Runtime.enable` CDP 泄漏**（rebrowser）² | — | ❌ 泄漏 | ❌ 泄漏 | ✅ **默认关闭** | +| 多 agent 共用**同一个**真实 Chrome、标签组隔离³ | ❌ 单 app | ⚠️ 共享 tab、无隔离 | ❌ 各开各的浏览器 | ✅ | | 权限面 | 16 个，含 `` | 完整 CDP | 完全控制 | **7 个，无 ``** | -> 一句话：**Claude in Chrome** 很好，但只能给 Claude 用；**web-access / 裸 `--remote-debugging-port`** 每次连接都会弹 Chrome 136+ 的同意框；**Playwright/Puppeteer/browser-use** 启的是**全新**浏览器（没登录、带自动化标记、常 headless → 被识破）。agent-browser-stealth 直接开你**已经登录好**的那个 Chrome，走原生消息扩展，在检测站上读起来就是个 100% 真人浏览器。证据见 [反检测](#反检测)。 +> 一句话：**Claude in Chrome** 很好，但只能给 Claude 用；**web-access / 裸 `--remote-debugging-port`** 每次连接都会弹 Chrome 136+ 的同意框；**Playwright/Puppeteer/browser-use** 启的是**全新**浏览器（没登录、带自动化标记、常 headless → 被识破）。agent-browser-stealth 直接开你**已经登录好**的那个 Chrome，走原生消息扩展，在检测站上读起来就是个 100% 真人浏览器。 +> +> ¹ 三家"真实 Chrome"工具在 CreepJS 上都 ~0%（毕竟是真浏览器），我们的是实测过的。² rebrowser `runtimeEnableLeak` —— 我们的中继路径实测无泄漏；Claude in Chrome 未独立测试（—）。³ web-access 也能跑并行子 agent，但无每会话隔离；本工具每个 `--session` 拿到自己彩色、命令隔离的标签组。实测数字见 [反检测](#反检测)。 ## 为什么要 fork