Give your AI agent eyes to see the entire internet. Read & search Twitter, Reddit, YouTube, GitHub, Bilibili, XiaoHongShu — one CLI, zero API fees.
Python agent-infrastructureai-agentai-searchautomationbilibiliclaude-codeclicursorfree-apillm-toolsmcppythonreddit-scrapertwitter-scraperweb-scraperxiaohongshuyoutube-transcript
docs: rewrite README (zh/en) for 1.6
Pnant5c68fb13301f
diff --git a/README.md b/README.mdindex 69462bfe..68641151 100644--- a/README.md+++ b/README.md@@ -6,7 +6,7 @@</p><p align="center">- 登录墙后的平台、视频里的内容、中文平台、干净的结构化数据,一条命令拿到,统一 JSON 输出+ 登录墙后的平台、视频里的内容、中文平台、干净的结构化数据,一条命令拿到</p><p align="center">@@ -21,11 +21,15 @@</p><p align="center">- <a href="#快速上手">快速开始</a> · <a href="docs/README_en.md">English</a> · <a href="#支持的平台">支持平台</a> · <a href="#设计理念">设计理念</a> · <a href="https://discord.gg/NSZz22R39">Discord</a>+ <a href="#快速上手">快速开始</a> · <a href="docs/README_en.md">English</a></p><p align="center">- <a href="https://discord.gg/NSZz22R39"><img src="https://img.shields.io/badge/Discord-Join-5865F2?logo=discord&logoColor=white" alt="加入 Agent Reach Discord 社区"></a>+ <a href="https://discord.gg/NSZz22R39"><img src="https://img.shields.io/badge/Discord-加入社区-5865F2?style=for-the-badge&logo=discord&logoColor=white" alt="加入 Agent Reach Discord 社区"></a>+</p>++<p align="center">+ 💬 有问题、想要新平台、或者想看看别人怎么用?来 <a href="https://discord.gg/NSZz22R39">Discord</a> 聊聊。</p>---@@ -34,11 +38,19 @@> [想出现在这里?](mailto:[email protected])-<details open>-<summary>点击折叠</summary>+<details>+<summary>点击展开</summary><table><tr>+<td width="180" align="center"><a href="https://get.brightdata.com/bdenai"><img src="docs/assets/sponsors/brightdata.png" alt="Bright Data 亮数据" width="150"></a></td>+<td><a href="https://get.brightdata.com/bdenai">Bright Data 亮数据</a> | 全球领先的数据采集基础设施。Scraper Studio 智能爬取器让你用自然语言一键生成爬虫,无需写代码;Unlocker 网站解锁器自动绕过反爬虫机制,轻松采集 Amazon、Google、LinkedIn、TikTok 等复杂网站数据。内置 350+ 现成 API 集成,覆盖电商、社交、搜索、地图等全品类数据源,直接调用开箱即用。1.5 亿+ 全球真实住宅 IP 保障采集稳定性。每月免费 5,000 次请求,<a href="https://get.brightdata.com/bdenai">立即免费注册</a>。</td>+</tr>+<tr>+<td width="180" align="center"><a href="https://www.rapidproxy.io/?ref=pann"><img src="docs/assets/sponsors/rapidproxy.png" alt="Rapidproxy" width="150"></a></td>+<td><a href="https://www.rapidproxy.io/?ref=pann">RapidProxy</a> 是专为开发者打造的数据采集代理解决方案,提供稳定可靠的住宅代理服务。拥有 9000 万+ 全球住宅 IP,覆盖 200+ 国家和地区,支持智能轮换与精准地理定位,可用于网页抓取、AI 数据采集与训练、SEO 监控、电商数据分析等场景,帮助提升数据采集效率与稳定性。支持 Playwright、Selenium、Puppeteer 等主流自动化框架,流量永久有效,价格低至 $0.55/GB,<a href="https://www.rapidproxy.io/?ref=pann">支持免费测试</a>。</td>+</tr>+<tr><td width="180" align="center"><a href="https://www.browseract.ai/Agent"><img src="docs/assets/sponsors/browseract.png" alt="BrowserAct" width="150"></a></td><td><a href="https://www.browseract.ai/Agent">BrowserAct</a> 支持从 Amazon、LinkedIn、X、Google Maps 等复杂网站提取你需要的任意数据。你只需用自然语言描述抓取需求,Agent 就会基于真实浏览器自动探索并测试页面流程,生成可靠、可复用的数据采集 Bot,并返回结构化结果。无需手动构建爬虫,无需编写代码。BrowserAct 内置隐身浏览、验证码处理和高质量住宅代理,帮助你更稳定地完成复杂网页数据采集。新用户注册即送 1000 积分,<a href="https://www.browseract.ai/Agent">立即免费试用</a>。</td></tr>@@ -62,14 +74,16 @@## 为什么需要 Agent Reach?+**Agent Reach 让你的 Agent 真正连上互联网,用上它自己够不着的平台和能力。**+Claude Code、Codex、Hermes Agent、OpenClaw 这类 CLI Agent 自带搜索和网页读取,但下面这些它们基本够不着:-- **需要浏览器会话的功能**:在小红书搜笔记、在 X 查账号动态、读职位详情等,按所选功能复用登录态。能公开读取的内容先用已有搜索和网页工具。-- **视频里的内容**:YouTube 和 B站的字幕,以及 YouTube 视频的语音转文字。+- **要登录才能用的功能**:在小红书搜笔记、在 X 查账号动态、读职位详情等。+- **视频里的内容**:B站和 YouTube 的字幕,以及 YouTube 视频、小宇宙播客的语音转文字。- **中文平台**:B站、小红书、V2EX、雪球,通用搜索覆盖很差。-- **干净的结构化数据**:雪球行情、V2EX 帖子和回复、GitHub 仓库和 issue,直接给字段,不用从 HTML 里猜。+- **干净的结构化数据**:雪球行情、V2EX 帖子和回复、GitHub 仓库和 issue,拿到的是整理好的内容。-Agent Reach 安装这些平台背后的工具,提供已测依赖与验收流程,然后给 Agent 一个统一入口:+Agent Reach 把这些平台背后的工具替你装好、测好,Agent 一条命令就能用:```bashagent-reach channels # 有哪些渠道、每个渠道能做什么@@ -77,43 +91,36 @@ agent-reach bilibili search "大模型" # agent-reach <渠道> <动作agent-reach read https://www.v2ex.com/t/123 # 按链接自动选渠道```-每条命令都输出同一种 JSON,Agent 不用学每个工具的用法,也不用解析各家五花八门的输出。--> `web` 渠道(网页搜索和阅读)只是保底:Agent 自带的搜索或网页读取不可用、或者读失败时再用它。-### ✅ 在你用之前,你可能想知道-- **适用于哪些 Agent?** 任何能执行命令行的 Agent。`agent-reach setup` 会放一份使用说明(skill,就是告诉 AI 怎么用 Agent Reach 的说明文件),Codex、OpenCode 等直接能读到;Claude Code、OpenClaw、Cursor 等装了的话,会在它们的 skill 目录里放一个指向这份说明的链接(相当于快捷方式)。-- **会碰我的浏览器和账号吗?** 查询默认不弹窗口,会先尝试复用本机已登录的网站,只读复制该站 Cookie,原浏览器不变;不读 Cookie 数据库或密码库。确实需要重新登录时才打开 Reach 专用窗口,由你本人完成。详见[会话配置](agent_reach/skill/references/sessions.md)。-- **支持哪些系统?** 优先保证 macOS,Linux 可用,Windows 尽力支持。-- **要我自己装别的软件吗?** `agent-reach add` 会准备 Node、OpenCLI、专用浏览器和插件;已有登录优先复用,确实需要时再登录。-- **怎么知道有新版本?** Agent Reach 每天最多在后台查一次,有新版本时 Agent 会用大白话告诉你改了什么、问你要不要更新,你同意了才更新。+- **适用于哪些 Agent?** 一切 Agent 都能用。+- **会碰我的浏览器和账号吗?** 要登录的平台用 Agent Reach 自己的专用浏览器,在后台运行、不弹窗。第一次用某个网站时,会把你 Chrome 里这个网站的登录复制一份过去,原浏览器不变,不读密码库。复制过来的通常是主账号,建议换成小号,做法见下面的"安全性"。详见[会话配置](agent_reach/skill/references/sessions.md)。---## 支持的平台-`agent-reach channels` 是实时的准确来源,下表是当前版本的情况。+下表是当前版本的情况。| 渠道 | 等级 | 能做什么 | 需要什么 ||---|---|---|---|-| github | 默认 | 搜仓库、读仓库/README、读 issue 和 PR(含评论) | 无(装了 `gh` 或配置 token 后额度更高) |-| youtube | 默认 | 视频信息、搜索、字幕、语音转文字 | 无;转文字需要 Groq 或 OpenAI Key |+| bilibili | 默认 | 搜索、视频信息、热门、字幕 | 无;字幕需要登录(扫码) || v2ex | 默认 | 热门、节点、帖子和回复、用户 | 无 |+| github | 默认 | 搜仓库、读仓库/README、读 issue 和 PR(含评论) | 无 |+| youtube | 默认 | 视频信息、搜索、字幕、语音转文字 | 无;转文字需要 Groq 或 OpenAI Key || web | 默认 | 网页搜索、网页阅读(保底) | 无 |-| bilibili | 默认 | 搜索、视频信息、热门、字幕 | 无;字幕需要登录(扫码) |-| xiaohongshu | 可选 | 搜笔记、读正文、看评论 | 先查公开笔记;站内与评论路径按需认证 |-| reddit | 可选 | 搜帖子、读正文、看评论、看版块 | 公开帖子和评论先用已有网页工具;浏览器功能按需登录 |-| linkedin | 可选 | 搜职位、读职位详情 | 先查公开网页;站内路径按实际结果决定是否登录 |-| boss | 可选 | 搜职位和 JD | 先查公开职位;站内路径按需认证 |+| xiaohongshu | 可选 | 搜笔记、读正文、看评论 | 登录 |+| xueqiu | 可选 | 实时行情、搜股票、热帖、热股 | 行情不用登录;热帖、热股要登录 |+| boss | 可选 | 搜职位和 JD | 登录 || xiaoyuzhou | 可选 | 播客转文字 | Groq Key |-| xueqiu | 可选 | 实时行情、搜股票、热帖、热股 | 行情免登录;热帖热股需要浏览器登录 |-| twitter | 实验 | 搜推文、用户资料、用户推文、读推文串 | 先查公开索引;动态和回复路径按需认证 |-| facebook | 实验 | 搜索、主页资料 | 先查公开索引;主页路径按需认证 |-| instagram | 实验 | 搜用户、主页资料、用户帖子 | 浏览器 + 登录 |--- **默认**:装好就能用;不想要也可以去掉(`agent-reach remove <渠道>`),以后更新不会再加回来。-- **可选**:用 `agent-reach add <渠道>` 按需开启,它需要的一切 Agent Reach 都会装到自己的目录里。+| linkedin | 可选 | 搜职位、读职位详情 | 公开职位不用登录;站内搜索要登录 |+| reddit | 可选 | 搜帖子、读正文、看评论、看版块 | 登录 |+| twitter | 实验 | 搜推文、用户资料、用户推文、读推文串 | 登录 |+| facebook | 实验 | 搜索、主页资料 | 登录 |+| instagram | 实验 | 搜用户、主页资料、用户帖子 | 登录 |++- **默认**:装好就能用;不想要也可以去掉,以后更新不会再加回来。+- **可选**:说一句"帮我加上 XX"就开启,需要的软件会自动装好。- **实验**:能用,但平台反爬更严,可能随时失效。---@@ -123,104 +130,74 @@ agent-reach read https://www.v2ex.com/t/123 # 按链接自动选渠道把这句话发给你的 Agent:```-帮我安装 Agent Reach,按 docs/install.md 优先复用已有登录态。+帮我安装 Agent Reach:https://raw.githubusercontent.com/Panniantong/Agent-Reach/main/docs/install.md```-剩下的 Agent 会照着 [install.md](docs/install.md) 自己做完。过程中你可能会被问到:+无论是安装还是升级,都只需要这一条命令就行了。++剩下的 Agent 自己做完。过程中你可能会被问到:- **允许它执行命令**:如果 Agent 说自己不能运行命令,去它的设置里打开"允许执行命令"。- **装 uv**:uv 是一个安装命令行程序的小工具,不需要管理员密码。电脑上没有的话 Agent 会先装它,会跟你说一声。-- **要不要加可选渠道**:装完后 Agent 会告诉你哪些已经能用(网页、YouTube、GitHub、V2EX、B站),以及还能加哪些。你选了再加,也可以以后随时说"帮我加上小红书",一条命令就装好。--需要你本人做的只有登录,Agent 从不替你登录,也不会要你的密码:+- **要不要加可选渠道**:装完后 Agent 会告诉你哪些已经能用(B站、V2EX、GitHub、YouTube、网页),以及还能加哪些。你选了再加,也可以以后随时说"帮我加上小红书",一条命令就装好。-- 小红书、Reddit、Twitter 等:先复用专用浏览器中已保存的登录;需要继承日常 Chrome 登录时先授权按站点导入。-- B站:字幕要扫码登录;不登录也能搜索和看视频信息。+关于登录:部分平台会直接复用你电脑上已有的登录。想换成某个特定账号(比如小号),跟 Agent 说一声就行。Agent 从不替你登录,也不会要你的密码。> ⚠️ PyPI 上叫 `agent-reach` 的包不是本项目,请勿从 PyPI 安装,只用 install.md 里的 GitHub 地址。-想自己动手也一样,照着 [install.md](docs/install.md) 敲命令就行。---```mermaid-flowchart LR- A[Agent] -->|agent-reach 渠道 动作| B[Agent Reach]- B --> C{按顺序尝试后端}- C -->|首选| D[上游工具 1]- C -->|失败时备选| E[上游工具 2]- D --> F[统一 JSON]- E --> F- F --> A-```----## 装好之后怎么管-直接用大白话跟 Agent 说就行,它会替你运行对应的命令:+Agent Reach 装的所有东西(各渠道的工具、登录、设置)都放在你电脑上同一个文件夹里(`~/.agent-reach/`),不散落到别处。所以升级、管理、删除都很省事,一句话就行。-| 你说 | Agent 运行 | 作用 |-|---|---|---|-| "看看 Agent Reach 状态" | `agent-reach doctor --text` | 每个渠道的状态:未开启 / 待配置 / 待登录 / 未验证 / 可用 / 故障,并标出"上次还好、现在坏了"的渠道 |-| "加上小红书" / "去掉小红书" | `agent-reach add` / `remove <渠道>` | 开启渠道并装好它需要的一切,你只需登录或提供 Key / 关闭渠道并清掉它的登录和专用工具(默认渠道也能去掉) |-| "B站登录过期了" / "退出登录" | `agent-reach login` / `logout <渠道>` | 登录(你本人操作)/ 清除登录 |-| "升级 Agent Reach" | `agent-reach update` | 一次更新程序、各渠道的工具(有已测版本的保持已测版本,其他用兼容范围内的最新版)和 skill,并告诉你还需要做什么,见 [update.md](docs/update.md) |-| "卸载 Agent Reach" | `agent-reach uninstall` | 见下面"安全性"里的卸载说明 |-| "让另一个 AI 也能用" | `agent-reach skill link <它的 skill 目录>` | 给不在自动名单里的 AI 也放一个说明链接 |+直接用大白话跟 Agent 说:-其他:`agent-reach check [渠道] --text` 真实跑一遍测试;`agent-reach configure <项>` 保存 `groq-key`、`openai-key`、`proxy`、`github-token`。+| 你说 | 它会做什么 |+|---|---|+| "看看 Agent Reach 状态" | 告诉你每个渠道现在能不能用,哪些是上次还好、现在坏了 |+| "加上小红书" / "去掉小红书" | 开启并装好它需要的一切 / 关掉它,清掉它的登录和专用工具 |+| "小红书登录过期了" / "换个账号" / "退出登录" | 打开登录窗口,由你本人登录 / 清掉这个网站的登录 |+| "升级 Agent Reach" | 一次更新程序和各渠道的工具,并告诉你还需要做什么 |+| "卸载 Agent Reach" | 先列出要删的东西,你确认后再删 |+| "让另一个 AI 也能用" | 给那个 AI 也放一份使用说明 |-出错时,JSON 里的 `error.code` 会告诉 Agent 下一步该做什么。每种错误和状态的处理办法见 [troubleshooting.md](docs/troubleshooting.md)。+出了问题,Agent 会告诉你是哪个平台、什么原因、下一步怎么做。常见问题见 [troubleshooting.md](docs/troubleshooting.md)。---## 设计理念-### 🔌 每个平台 = 首选 + 备选的有序后端列表--每个渠道的每个动作都有一个有序的后端列表。比如 `bilibili read` 先用 bili-cli,失败了自动换 B站公开接口;`web read` 先用 Jina Reader,再用 Exa。返回的 JSON 里 `backend` 字段会写明这次实际用的是谁。+### 🔌 每个平台 = 首选 + 备选的有序工具列表-渠道已经加上时,搜索和按链接阅读直接走这个渠道。还没加的渠道,带 `public_web` 声明的关键词搜索和按链接阅读会先复用网页搜索/读取:成功就直接交付,不安装浏览器或要求登录;结果来自公开网页,完整性与时效需核对。公开路径不足再尝试站内工具,实际遇到登录墙才提示登录。`--backend public-web` 只查公开内容;`--backend opencli` 等声明后端直接使用站内能力。评论、字幕和账号列表保留各自的动作要求。+每个平台的每个动作都有一个按顺序尝试的工具列表。比如读 B站视频先用 bili-cli,失败了自动换 B站公开接口;读网页先用 Jina Reader,再用 Exa。-平台改版、工具失效是常态。换后端只是调整这个列表,Agent 的调用方式不变。--### 📌 已测版本,私有安装--Agent Reach 装的工具位于 `~/.agent-reach/tools/`,可复用现有 OpenCLI。有已测版本的工具按已测版本安装,`update` 也不会改动;其他工具用兼容范围内的最新版;Boss 固定到已验证提交;`doctor` 显示实际版本。--### 🩺 能用要靠实测--`doctor` 只做离线检查;`check` 会真的去平台上请求一次。一个渠道只有通过实测才会标为"可用"。--### 统一输出--所有 Agent 命令都输出一个 JSON:--```json-{"schema_version": 1, "ok": true, "channel": "v2ex", "action": "hot", "backend": "v2ex-api",- "items": [{"title": "...", "url": "...", "text": "..."}], "meta": {}, "error": null}+```mermaid+flowchart LR+ A[Agent] -->|agent-reach 渠道 动作| B[Agent Reach]+ B --> C{按顺序尝试}+ C -->|首选| D[上游工具 1]+ C -->|失败时备选| E[上游工具 2]+ D --> F[结果]+ E --> F+ F --> A```-失败时 `ok` 为 `false`,`error` 里有 `code`(`NEED_SETUP` / `NEED_LOGIN` / `UPSTREAM_BROKEN` / `TIMEOUT` / `RATE_LIMITED` / `NOT_FOUND` / `BAD_INPUT`)、`message` 和给 Agent 的 `hint`。+平台改版、工具失效是常态。换工具只是调整这个列表,Agent 的用法不变。---## 安全性-- **所有东西都在 `~/.agent-reach/` 下**:`config.yaml`(Key 和代理)、`state.json`、`tools/`(私有工具)、`sessions/`(命令行工具的登录状态)、`browser/`(专用浏览器的资料目录)。唯一例外是使用说明:一份放在 `~/.agents/skills/agent-reach/`,其他 AI 的 skill 目录里只放指向它的链接。-- **从不替你登录,不默认搬运 Cookie。** 先复用选定会话;需要认证时才进入明确选择的工具登录流程。Boss 原生 CDP 同步按工具机制保存到私有目录。-- **CDP 调试端口只监听本机**。Boss 标准 CDP 窗口需用户使用后自行关闭;旧扩展专用模式的空闲退出机制不适用于它。-- **建议用小号登录。** 平台可能限制自动化访问的账号,别拿主账号冒险。-- **API Key 不要发在聊天里**:在你自己的终端运行 `agent-reach configure groq-key`,会隐藏输入。+- **只读**:只搜索和读取,不发帖、不评论、不点赞、不私信。+- **从不替你登录,也不要你的密码。** 部分平台会复用你电脑上已有的登录(只复制这一个网站,原浏览器不变);确实需要重新登录时才打开专用窗口,由你本人完成。+- **专用浏览器只对你这台电脑开放**,别的电脑连不上它。+- **建议用小号登录。** 复用过来的通常是主账号,而平台可能限制自动访问的账号。想换成小号,跟 Agent 说一声就行。+- **Key 不要发在聊天里**:在你自己的终端里运行 `agent-reach configure groq-key` 填写。### 🗑️ 卸载-```bash-agent-reach uninstall --dry-run # 先看会删哪些东西-agent-reach uninstall # 删除 Agent Reach 装的工具、登录、配置和 skill-```--卸载前先关掉 Agent Reach 专用浏览器。命令最后会告诉你怎么卸载程序本身(比如 `uv tool uninstall agent-reach`)。+跟 Agent 说"卸载 Agent Reach"。它会先列出要删的东西,你确认后再删。只删 Agent Reach 自己装的东西,你的其他文件和浏览器都不动。最后它会告诉你怎么删掉程序本身。---@@ -232,8 +209,6 @@ agent-reach uninstall # 删除 Agent Reach 装的工具、登录、- 每个渠道我会尽量保证**能用、好用、免费**- 平台改了反爬或者 API 变了,我会想办法解决-为 Web 4.0 基建贡献一份自己的力量。-Star 一下,下次需要的时候能找到。⭐---@@ -247,19 +222,9 @@ Star 一下,下次需要的时候能找到。⭐- 📧 **Email:** [email protected]- 🐦 **Twitter/X:** [@Neo_Reidlab](https://x.com/Neo_Reidlab)-## 业务合作 / Agent 落地--我正在承接 Agent 相关的定制与落地合作。--如果你在企业生产、运营、市场、投研、数据处理、内容处理或其他业务流程里,有希望用 Agent 自动化的环节,欢迎加我微信交流。--不需要你已经想清楚方案。只要你有真实流程、真实问题或真实需求,我可以一起判断 Agent 能不能解决、怎么做。--加好友请备注:`业务 + 你想让 Agent 帮你做什么`+## 业务合作 / 交流群-Builder 也欢迎备注:`Builder + 你在做什么`--只是想进交流群,备注:`加群`+有相关合作,或者想加交流群,加我微信:<p align="center"><img src="docs/wechat-group-qr.jpg" width="280" alt="WeChat QR">@@ -267,6 +232,10 @@ Builder 也欢迎备注:`Builder + 你在做什么`> Bug 反馈和功能请求请用 [GitHub Issues](https://github.com/Panniantong/Agent-Reach/issues),更容易跟踪。+## 防骗提醒++Agent Reach 没有官方代币、币、投资产品、领取手续费活动、钱包连接,也没有 Solana/Pump.fun 项目。任何借用 Agent Reach 名字、GitHub 地址或作者身份的加密项目都与本仓库无关,请勿据此连接钱包或领取任何费用。+## License[MIT](LICENSE)@@ -280,9 +249,3 @@ Builder 也欢迎备注:`Builder + 你在做什么`## Star History<a href="https://www.star-history.com/?type=date&repos=Panniantong%2FAgent-Reach"><picture><source media="(prefers-color-scheme: dark)" srcset="https://api.star-history.com/chart?repos=Panniantong/Agent-Reach&type=date&theme=dark&legend=top-left&sealed_token=K3_u-LJQTVURYu-38Tqa_VWJOSqMf_HbAw-QKSdGwEq6seqznugdIpXdSeztEdOutT40IBXwVxTmg8wS_OSygb5UWf1x8e-Fai6aygrjq6QH8vU09EcqQCN7atp-76HmxX-j9fnZ9NiSrLDNzK98TnXBFJ_Wb_y80I0nWr3O8DdGnLXFhAJgoNK3Jz8D" /><source media="(prefers-color-scheme: light)" srcset="https://api.star-history.com/chart?repos=Panniantong/Agent-Reach&type=date&legend=top-left&sealed_token=P2746KOq7grpS8Q-nrqEzci1-Z0-dOw-M3KEqju-l3TyF24NMyRDR7TnxdReJWlXyomoT4mjjqC28-2-c2G6CnzmS1hgYdEDiGPmLkmEqKP5tgjORXshdrUFxoSxTqmIKEMFFmGZUX1v3ec-q_XMyftTVWzluiQH7CvKoZ1uDKU3PJN05mO22u7qlLeG" /><img alt="Star History Chart" src="https://api.star-history.com/chart?repos=Panniantong/Agent-Reach&type=date&legend=top-left&sealed_token=P2746KOq7grpS8Q-nrqEzci1-Z0-dOw-M3KEqju-l3TyF24NMyRDR7TnxdReJWlXyomoT4mjjqC28-2-c2G6CnzmS1hgYdEDiGPmLkmEqKP5tgjORXshdrUFxoSxTqmIKEMFFmGZUX1v3ec-q_XMyftTVWzluiQH7CvKoZ1uDKU3PJN05mO22u7qlLeG" /></picture></a>----### 共用专用浏览器 / Shared dedicated browser--浏览器渠道共用一个 Reach 配置:CloakBrowser 优先,下载失败回退本机 Chrome;闲置 10 分钟关闭,下次自动重开。渠道数据由 OpenCLI / 原生 CLI 提取。授权登录态继承、自动安装插件和按站点清理见 [浏览器说明](docs/cloakbrowser.md)。小宇宙转录保留为可选渠道,需要 Groq Key。diff --git a/docs/README_en.md b/docs/README_en.mdindex 436b0730..0d3be439 100644--- a/docs/README_en.md+++ b/docs/README_en.md@@ -6,7 +6,7 @@</p><p align="center">- Login-walled platforms, what's inside videos, Chinese platforms, clean structured data. One command, one JSON shape.+ Login-walled platforms, what's inside videos, Chinese platforms, clean structured data. One command.</p><p align="center">@@ -21,14 +21,16 @@</p><p align="center">- <a href="#quick-start">Quick Start</a> · <a href="../README.md">中文</a> · <a href="#supported-platforms">Platforms</a> · <a href="#design-philosophy">Philosophy</a> · <a href="https://discord.gg/NSZz22R39">Discord</a>+ <a href="#quick-start">Quick Start</a> · <a href="../README.md">中文</a></p><p align="center">- <a href="https://discord.gg/NSZz22R39"><img src="https://img.shields.io/badge/Discord-Join-5865F2?logo=discord&logoColor=white" alt="Join the Agent Reach Discord community"></a>+ <a href="https://discord.gg/NSZz22R39"><img src="https://img.shields.io/badge/Discord-Join%20us-5865F2?style=for-the-badge&logo=discord&logoColor=white" alt="Join the Agent Reach Discord community"></a></p>-> **No token or crypto affiliation:** Agent Reach has no official token, coin, investment product, fee-claim program, wallet connection, or Solana/Pump.fun project. Any crypto project using the Agent Reach name, GitHub URL, or author identity is not affiliated with this repository. Do not connect a wallet or claim fees based on messages, posts, or links that say otherwise.+<p align="center">+ 💬 Questions, a platform you want added, or curious how others use it? Come chat on <a href="https://discord.gg/NSZz22R39">Discord</a>.+</p>---@@ -36,11 +38,19 @@> [Want to appear here?](mailto:[email protected])-<details open>-<summary>Click to collapse</summary>+<details>+<summary>Click to expand</summary><table><tr>+<td width="180" align="center"><a href="https://get.brightdata.com/bdenai"><img src="assets/sponsors/brightdata.png" alt="Bright Data" width="150"></a></td>+<td><a href="https://get.brightdata.com/bdenai">Bright Data</a> | The world's leading web data infrastructure. Scraper Studio builds a scraper from a plain-language description, no code needed; Web Unlocker gets past anti-bot defenses so you can collect data from hard sites like Amazon, Google, LinkedIn and TikTok. 350+ ready-made API integrations cover e-commerce, social, search, maps and more, ready to call out of the box. 150M+ real residential IPs worldwide keep collection stable. 5,000 free requests every month. <a href="https://get.brightdata.com/bdenai">Sign up free</a>.</td>+</tr>+<tr>+<td width="180" align="center"><a href="https://www.rapidproxy.io/?ref=pann"><img src="assets/sponsors/rapidproxy.png" alt="Rapidproxy" width="150"></a></td>+<td><a href="https://www.rapidproxy.io/?ref=pann">RapidProxy</a> is a data-collection proxy service built for developers, with stable, reliable residential proxies. 90M+ residential IPs worldwide cover 200+ countries and regions, with smart rotation and precise geo-targeting, for web scraping, AI data collection and training, SEO monitoring, e-commerce analytics and more. Works with Playwright, Selenium, Puppeteer and other popular automation frameworks. Traffic never expires, prices start at $0.55/GB, and <a href="https://www.rapidproxy.io/?ref=pann">free trials are available</a>.</td>+</tr>+<tr><td width="180" align="center"><a href="https://www.browseract.ai/Agent"><img src="assets/sponsors/browseract.png" alt="BrowserAct" width="150"></a></td><td><a href="https://www.browseract.ai/Agent">BrowserAct</a> extracts any data you need from complex websites such as Amazon, LinkedIn, X, and Google Maps. Simply describe your extraction request in natural language, and its Agent will explore and test page flows in a real browser, generate a reliable reusable data-collection Bot, and return structured results. There is no need to build a scraper or write code. Built-in stealth browsing, CAPTCHA handling, and high-quality residential proxies help make complex web data extraction more reliable. New users receive 1,000 credits upon registration. <a href="https://www.browseract.ai/Agent">Try it free now</a>.</td></tr>@@ -64,59 +74,53 @@## Why Agent Reach?+**Agent Reach connects your agent to the internet for real, giving it the platforms and abilities it can't reach on its own.**+CLI agents such as Claude Code, Codex, Hermes Agent and OpenClaw come with web search and page fetching, but they mostly can't reach:-- **Features that need a browser session**: searching XiaoHongShu notes, listing X account posts or reading job details. Reuse an authorized session for the selected action; public content stays on existing search and web tools.-- **What's inside video**: YouTube and Bilibili subtitles, plus speech-to-text for YouTube videos.+- **Features that need a login**: listing an X account's posts, searching Reddit, reading LinkedIn job details or XiaoHongShu notes.+- **What's inside video**: YouTube and Bilibili subtitles, plus speech-to-text for YouTube videos and Xiaoyuzhou podcasts.- **Chinese platforms**: Bilibili, XiaoHongShu, V2EX and Xueqiu, which general search covers poorly.-- **Clean structured data**: Xueqiu quotes, V2EX topics and replies, GitHub repos and issues, as fields instead of HTML to guess from.+- **Clean structured data**: Xueqiu quotes, V2EX topics and replies, GitHub repos and issues, already cleaned up.-Agent Reach installs the open-source tools behind these platforms, keeps them up to date, health-checks them, and gives the agent one entry point:+Agent Reach installs and tests the tools behind these platforms for you, so your agent can use them with one command:```bashagent-reach channels # which channels exist and what each can do-agent-reach bilibili search "LLM" # agent-reach <channel> <action> [args]+agent-reach youtube search "LLM agents" # agent-reach <channel> <action> [args]agent-reach read https://www.v2ex.com/t/123 # picks the channel from the link```-Every command prints the same JSON shape, so the agent doesn't have to learn each tool or parse each tool's output.--> The `web` channel (web search and page reading) is only a fallback, for when the agent's own search or fetch is missing or fails.-### ✅ Before you start, you might want to know-- **Which agents?** Any agent that can run shell commands. `agent-reach setup` places one copy of the usage instructions (a skill) that Codex, OpenCode and others read directly, and links it into the skill folders of Claude Code, OpenClaw, Cursor and others when they are installed.-- **Does it cost money?** The program and its backends are open source and free. The only keys you may need are for speech-to-text: Groq (has a free tier) or OpenAI Whisper.-- **Sessions**: browser channels share the Reach-owned profile. Previously saved Reach sessions are reused; importing a site login from your daily Chrome requires explicit authorization.-- **Which OS?** macOS first, Linux works, Windows best effort.-- **Dependencies**: add prepares Node, OpenCLI, the dedicated browser and extension. Log in only when the requested action requires it.-- **How do I hear about new versions?** Agent Reach checks at most once a day in the background. When there is a new version, your agent tells you in plain words what changed and asks whether to update; it updates only if you say yes.+- **Which agents?** Every agent.+- **Does it touch my browser and accounts?** Login channels use Agent Reach's own browser, in the background without windows. The first time a site is used, that site's login is copied over from your Chrome; your browser is not changed and passwords are never read. The copied login is usually your main account, so switching to a secondary one is recommended, see Security below.---## Supported Platforms-`agent-reach channels` is the live source of truth; this table reflects the current release.+This table reflects the current release.| Channel | Level | What it does | Needs ||---|---|---|---|-| github | default | search repos, read a repo/README, read issues and PRs with comments | nothing (higher limits with `gh` or a token) |+| github | default | search repos, read a repo/README, read issues and PRs with comments | nothing || youtube | default | video info, search, subtitles, speech-to-text | nothing; speech-to-text needs a Groq or OpenAI key |-| v2ex | default | hot topics, nodes, topics with replies, users | nothing || web | default | web search, page reading (fallback) | nothing |+| v2ex | default | hot topics, nodes, topics with replies, users | nothing || bilibili | default | search, video info, trending, subtitles | nothing; subtitles need a login (QR scan) |-| xiaohongshu | optional | search notes, read text, view comments | browser + login |-| reddit | optional | search posts, read text/comments, browse a subreddit | public posts/comments use existing web tools first; browser actions request login only if needed |-| linkedin | optional | job search | login |-| boss | optional | job and JD search | browser + login |-| xiaoyuzhou | Optional | Podcast transcription | Groq key |-| xueqiu | optional | real-time quotes, stock search, hot posts, hot stocks | quotes need nothing; hot posts/stocks need a browser login |-| twitter | experimental | search tweets, profiles, user tweets, threads | browser + login |-| facebook | experimental | search, profiles | browser + login |-| instagram | experimental | user search, profiles, user posts | browser + login |--- **default**: works right after install; you can remove one (`agent-reach remove <channel>`) and updates won't add it back.-- **optional**: turn on with `agent-reach add <channel>`; Agent Reach installs everything it needs into its own folder.+| reddit | optional | search posts, read text/comments, browse a subreddit | login |+| linkedin | optional | search jobs, read job details | public jobs need nothing; in-site search needs a login |+| xiaohongshu | optional | search notes, read text, view comments | login |+| xueqiu | optional | real-time quotes, stock search, hot posts, hot stocks | quotes need nothing; hot posts and stocks need a login |+| boss | optional | job and JD search | login |+| xiaoyuzhou | optional | podcast transcription | Groq key |+| twitter | experimental | search tweets, profiles, user tweets, threads | login |+| facebook | experimental | search, profiles | login |+| instagram | experimental | user search, profiles, user posts | login |++- **default**: works right after install; you can remove one, and updates won't add it back.+- **optional**: say "add XX" to turn it on; whatever it needs is installed for you.- **experimental**: works, but the platform fights automation harder and it can break at any time.---@@ -126,101 +130,86 @@ Every command prints the same JSON shape, so the agent doesn't have to learn eacSend this to your agent:```-Install this modified Agent Reach checkout/wheel using its docs/install.md.+Install Agent Reach: https://raw.githubusercontent.com/Panniantong/Agent-Reach/main/docs/install.md```-The agent does the rest by following [install.md](install.md). Along the way you may be asked to:+Whether you're installing or upgrading, this one command is all you need.++The agent does the rest. Along the way you may be asked to:- **Let it run commands**: if the agent says it can't run commands, turn on command execution in its settings.- **Allow installing uv**: uv is a small tool that installs command-line programs; no admin password needed. The agent installs it first if it's missing and tells you.- **Pick optional channels**: afterwards the agent tells you what works now (web, YouTube, GitHub, V2EX, Bilibili) and what else you can add. Pick now, or later just say "add Xiaohongshu"; one command sets it up.-The only thing you do yourself is log in; the agent never logs in for you and never asks for your password:--- Xiaohongshu, Reddit, Twitter: reuse the shared dedicated browser; import a selected site login only with explicit authorization.-- Bilibili: subtitles need a QR-scan login; search and video info work without it.+About logins: some platforms reuse the logins already on your computer. Want a specific account instead, such as a secondary one? Just tell your agent. The agent never logs in for you and never asks for your password.> ⚠️ The PyPI package named `agent-reach` is NOT this project. Don't install it from PyPI; use the GitHub URL in install.md.-Prefer doing it by hand? Run the commands in [install.md](install.md) yourself.--```mermaid-flowchart LR- A[Agent] -->|agent-reach channel action| B[Agent Reach]- B --> C{try backends in order}- C -->|preferred| D[upstream tool 1]- C -->|fallback on failure| E[upstream tool 2]- D --> F[one JSON envelope]- E --> F- F --> A-```----## After Install: Managing It-Just tell your agent in plain words; it runs the command for you:+Everything Agent Reach installs (each channel's tools, logins and settings) lives in one folder on your computer (`~/.agent-reach/`) instead of being scattered around. That makes upgrading, managing and removing it easy: one sentence does it.-| You say | Agent runs | What it does |-|---|---|---|-| "How is Agent Reach doing?" | `agent-reach doctor --text` | status of every channel: not_added / needs_setup / needs_login / unverified / ready / broken, and which ones broke since the last check |-| "Add Xiaohongshu" / "Remove Xiaohongshu" | `agent-reach add` / `remove <channel>` | turn a channel on and install everything it needs; you only log in or provide a key / turn it off and clear its login and private tools (default channels too) |-| "My Bilibili login expired" / "Log out" | `agent-reach login` / `logout <channel>` | log in (you do it) / forget the login |-| "Upgrade Agent Reach" | `agent-reach update` | update the program, every channel's tools (tested versions stay put; others get the newest supported version) and the skill in one go, and tell you anything left to do, see [update.md](update.md) |-| "Uninstall Agent Reach" | `agent-reach uninstall` | see Uninstall under Security below |-| "Let my other AI use it too" | `agent-reach skill link <its skills folder>` | link the instructions for an AI not covered automatically |+Just tell your agent in plain words:-Also: `agent-reach check [channel] --text` runs live tests; `agent-reach configure <key>` saves `groq-key`, `openai-key`, `proxy`, `github-token`.+| You say | What it does |+|---|---|+| "How is Agent Reach doing?" | tells you which channels work right now, and which worked before but broke since |+| "Add Xiaohongshu" / "Remove Xiaohongshu" | turns it on and installs what it needs / turns it off and clears its login and private tools |+| "My Xiaohongshu login expired" / "Switch accounts" / "Log out" | opens a login window where you sign in yourself / clears that site's login |+| "Upgrade Agent Reach" | updates the program and every channel's tools in one go, and tells you anything left to do |+| "Uninstall Agent Reach" | lists what would be removed first, and removes it once you confirm |+| "Let my other AI use it too" | gives that AI its own copy of the instructions |-When something fails, `error.code` in the JSON tells the agent what to do next. Every error code and status is explained in [troubleshooting.md](troubleshooting.md).+When something goes wrong, your agent tells you which platform, why, and what to do next. Common problems: [troubleshooting.md](troubleshooting.md).---## Design Philosophy-### 🔌 Every platform = an ordered backend list (primary + fallbacks)--Each action of each channel has an ordered list of backends. `bilibili read` tries bili-cli first and falls back to Bilibili's public API; `web read` tries Jina Reader, then Exa. The `backend` field in the JSON says which one actually answered.--Platforms change and tools break all the time. Switching backends only reorders that list; the way the agent calls Agent Reach stays the same.+### 🔌 Every platform = an ordered list of tools (primary + fallbacks)-### 📌 Tested versions, private installs+Each action on each platform has an ordered list of tools to try. Reading a Bilibili video tries bili-cli first and falls back to Bilibili's public API; reading a web page tries Jina Reader, then Exa.-Tools are private under ~/.agent-reach/tools/. Tools with a tested version install at that version, and `update` keeps them there; other tools use the newest version inside their supported range; Boss is pinned to a verified Git commit. `doctor` shows the installed versions.+```mermaid+flowchart LR+ A[Agent] -->|agent-reach channel action| B[Agent Reach]+ B --> C{try in order}+ C -->|preferred| D[upstream tool 1]+ C -->|fallback on failure| E[upstream tool 2]+ D --> F[result]+ E --> F+ F --> A+```-### 🩺 Working means live-tested+Platforms change and tools break all the time. Switching tools only reorders that list; the way your agent uses Agent Reach stays the same.-`doctor` only checks offline; `check` makes a real request to the platform. A channel is marked `ready` only after it passes a live check.+----### One output shape+## Security-Every agent command prints one JSON object:+- **Read-only**: it searches and reads; it never posts, comments, likes or sends messages.+- **It never logs in for you or asks for your password.** Some platforms reuse a login already on your computer (that one site only, your browser unchanged); a login window opens only when you really need to sign in again, and you do it yourself.+- **The dedicated browser is open to this computer only**; other machines can't connect to it.+- **Use a secondary account.** The reused login is usually your main account, and platforms may restrict accounts used for automation. To switch, just tell your agent.+- **Never paste keys into a chat**: run `agent-reach configure groq-key` in your own terminal to enter one.-```json-{"schema_version": 1, "ok": true, "channel": "v2ex", "action": "hot", "backend": "v2ex-api",- "items": [{"title": "...", "url": "...", "text": "..."}], "meta": {}, "error": null}-```+### 🗑️ Uninstall-On failure `ok` is `false` and `error` holds a `code` (`NEED_SETUP` / `NEED_LOGIN` / `UPSTREAM_BROKEN` / `TIMEOUT` / `RATE_LIMITED` / `NOT_FOUND` / `BAD_INPUT`), a `message`, and a `hint` for the agent.+Tell your agent "Uninstall Agent Reach". It lists what would be removed first and removes it once you confirm. Only what Agent Reach installed is removed; your other files and browsers stay. At the end it tells you how to remove the program itself.----## Security--- **Everything lives under `~/.agent-reach/`**: `config.yaml` (keys and proxy), `state.json`, `tools/` (private tool installs), `sessions/` (logins of CLI tools), `browser/` (the dedicated browser profile). The one exception is the skill: one copy in `~/.agents/skills/agent-reach/`, with only links to it in other AIs' skill folders.-- **You complete login yourself.** Saved Reach sessions are reused. With your explicit permission, Reach can copy one site's cookies from a connected Chrome profile; it never reads Chrome's cookie database.-- **CDP** listens on loopback only. BOSS shares the owned browser, which closes after ten idle minutes.-- **Use a secondary account for login channels.** Platforms may restrict accounts used for automated access.-- **Don't paste API keys into chat**: run `agent-reach configure groq-key` in your own terminal; input is hidden.+## ⭐ Why star it-### 🗑️ Uninstall+I use this project every day myself, so I'll keep maintaining it.-```bash-agent-reach uninstall --dry-run # see what would be removed-agent-reach uninstall # remove the tools, logins, config and skill Agent Reach installed-```+- When people ask for a new platform, I'll keep adding them+- I try to keep every channel **working, easy to use and free**+- When a platform changes its anti-bot rules or API, I'll find a way around it-Close the Agent Reach browser first. The output ends with the command that removes the program itself (e.g. `uv tool uninstall agent-reach`).+Star it so you can find it next time you need it. ⭐---@@ -233,7 +222,9 @@ Close the Agent Reach browser first. The output ends with the command that remov- 📧 **Email:** [email protected]- 🐦 **Twitter/X:** [@Neo_Reidlab](https://x.com/Neo_Reidlab)-For collaboration or questions, add me on WeChat — I'll invite you to the community group:+## Collaboration / community group++For collaboration, or to join the community group, add me on WeChat:<p align="center"><img src="wechat-group-qr.jpg" width="280" alt="WeChat QR">@@ -241,6 +232,10 @@ For collaboration or questions, add me on WeChat — I'll invite you to the comm> For bug reports and feature requests, please use [GitHub Issues](https://github.com/Panniantong/Agent-Reach/issues) — easier to track.+## No token or crypto affiliation++Agent Reach has no official token, coin, investment product, fee-claim program, wallet connection, or Solana/Pump.fun project. Any crypto project using the Agent Reach name, GitHub URL, or author identity is not affiliated with this repository. Do not connect a wallet or claim fees based on messages, posts, or links that say otherwise.+## License[MIT](../LICENSE)@@ -254,9 +249,3 @@ For collaboration or questions, add me on WeChat — I'll invite you to the comm## Star History<a href="https://www.star-history.com/?type=date&repos=Panniantong%2FAgent-Reach"><picture><source media="(prefers-color-scheme: dark)" srcset="https://api.star-history.com/chart?repos=Panniantong/Agent-Reach&type=date&theme=dark&legend=top-left&sealed_token=K3_u-LJQTVURYu-38Tqa_VWJOSqMf_HbAw-QKSdGwEq6seqznugdIpXdSeztEdOutT40IBXwVxTmg8wS_OSygb5UWf1x8e-Fai6aygrjq6QH8vU09EcqQCN7atp-76HmxX-j9fnZ9NiSrLDNzK98TnXBFJ_Wb_y80I0nWr3O8DdGnLXFhAJgoNK3Jz8D" /><source media="(prefers-color-scheme: light)" srcset="https://api.star-history.com/chart?repos=Panniantong/Agent-Reach&type=date&legend=top-left&sealed_token=P2746KOq7grpS8Q-nrqEzci1-Z0-dOw-M3KEqju-l3TyF24NMyRDR7TnxdReJWlXyomoT4mjjqC28-2-c2G6CnzmS1hgYdEDiGPmLkmEqKP5tgjORXshdrUFxoSxTqmIKEMFFmGZUX1v3ec-q_XMyftTVWzluiQH7CvKoZ1uDKU3PJN05mO22u7qlLeG" /><img alt="Star History Chart" src="https://api.star-history.com/chart?repos=Panniantong/Agent-Reach&type=date&legend=top-left&sealed_token=P2746KOq7grpS8Q-nrqEzci1-Z0-dOw-M3KEqju-l3TyF24NMyRDR7TnxdReJWlXyomoT4mjjqC28-2-c2G6CnzmS1hgYdEDiGPmLkmEqKP5tgjORXshdrUFxoSxTqmIKEMFFmGZUX1v3ec-q_XMyftTVWzluiQH7CvKoZ1uDKU3PJN05mO22u7qlLeG" /></picture></a>----### Shared dedicated browser--Browser channels share one Reach profile. CloakBrowser is preferred, with installed Chrome as the download-failure fallback. It closes after ten idle minutes and reopens on demand. OpenCLI or native upstream CLIs extract the data. See [browser setup](cloakbrowser.md) for login reuse, automatic extension loading and site-scoped cleanup. Xiaoyuzhou transcription remains optional and requires a Groq key.diff --git a/docs/assets/sponsors/brightdata.png b/docs/assets/sponsors/brightdata.pngnew file mode 100644index 00000000..5de42dabBinary files /dev/null and b/docs/assets/sponsors/brightdata.png differdiff --git a/docs/assets/sponsors/rapidproxy.png b/docs/assets/sponsors/rapidproxy.pngnew file mode 100644index 00000000..ad0c3a3fBinary files /dev/null and b/docs/assets/sponsors/rapidproxy.png differdiff --git a/docs/troubleshooting.md b/docs/troubleshooting.mdindex c44afde4..06376960 100644--- a/docs/troubleshooting.md+++ b/docs/troubleshooting.md@@ -56,7 +56,7 @@ Search-only acceptance is shown as search_ready, not proof that every action wor## Install, update and roll back-Agent Reach is installed with uv only. If `agent-reach update` says the program was installed another way, reinstall once as in [install.md](install.md); your channels, keys and logins under `~/.agent-reach/` are kept.+Agent Reach is installed with uv. A copy installed another way (pip or pipx, as 1.5 was) moves itself to uv on the next `agent-reach update`; your channels, keys and logins under `~/.agent-reach/` are kept.A new version broke something? Go back to the previous release, then line the tools and skill up with it: