[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"$f35cq5kbwsc5f3":3},{"code":4,"message":5,"data":6,"timestamp":54},10000,"success",{"full_name":7,"position":8,"description":9,"language":10,"stars":11,"forks":12,"stars_gained":13,"age_days":14,"hn_threads":15,"one_liner":27,"summary_md":28,"topic":29,"github_topics":32,"license":50,"homepage":51,"updated_at":52,"history":53},"Panniantong\u002FAgent-Reach",10,"Give your AI agent eyes to see the entire internet. Read & search Twitter, Reddit, YouTube, GitHub, Bilibili, XiaoHongShu — one CLI, zero API fees.","Python",101655,8975,16814,242,[16,21],{"id":17,"title":18,"points":19,"comments":19,"posted_at":20},49207806,"Agent Reach: An open-source CLI that gives AI agents access to the internet",4,"2026-08-07",{"id":22,"title":23,"points":24,"comments":25,"posted_at":26},49146136,"Anyone used Agent Reach yet? any good?",1,0,"2026-08-02","给 AI Agent 一条命令访问登录墙、视频与中文平台的读取搜索 CLI。","## 解决什么问题\n\nClaude Code、Codex 这类 CLI Agent 自带网页搜索和阅读，但很多有价值的内容它够不着：小红书笔记、X 账号动态、B站和 YouTube 的字幕、雪球行情、V2EX 的帖子与回复。这些要么在登录墙后面，要么在视频音频里，通用搜索覆盖很差。Agent Reach 把这些平台背后的抓取工具、会话和降级逻辑打包，Agent 用一条 `agent-reach \u003C渠道> \u003C动作>` 就能取数，不必自己逐个拼装。\n\n## 有什么不同\n\nMCP 服务器、LangChain 一类连接器给的是协议，具体站点集成还得自己写、自己维护；Bright Data 这类采集服务能覆盖同样平台，但走的是付费托管。Agent Reach 的关键设计是每个平台每个动作配一个有序工具列表：先试首选工具，失败自动换备选（如先 bili-cli 再走 B 站公开接口），平台改版时改的是列表，Agent 的用法不变。另外它把渠道分成默认、可选、实验三档，登录态是从你 Chrome 复制单个站点的会话到独立后台浏览器，而不是驱动你日常用的浏览器。\n\n## 适合谁\n\n在本地跑 CLI Agent、经常需要 X、Reddit、YouTube、B站、小红书、V2EX 内容的个人开发者和分析师，适合调研、监控、内容聚合这类单机场景。不适合需要稳定、大批量、有合同保障的数据管道的团队——反爬变化和账号封禁正是这套方案要人工消化的问题。需要发帖、评论、点赞等写操作的项目也不适用，它只做只读。\n\n## 上手成本\n\n需要 Python 3.10+ 和 uv，所有内容装在 `~\u002F.agent-reach\u002F` 一个目录里。安装由 Agent 按 GitHub 上的 install.md 完成；注意 README 明确提醒 PyPI 上同名包并非本项目。可选渠道会额外装依赖，YouTube 和小宇宙转文字需要 Groq 或 OpenAI Key。真正的风险在运维：Twitter、Facebook、Instagram 属实验档，作者自己也承认随时可能失效；复用来的登录通常是主账号，README 建议换小号，因为平台可能限制自动访问。MIT 协议。\n\n## 点评\n\n项目 2026 年 2 月才创建，月榜 Python 第 10、本月 +16,814 star，说明这是一次近期陡升而非长期积累；Hacker News 上只有一条 4 分的帖子，热度基本来自 GitHub 自然发现。按平台维护降级列表的思路务实，但这套设计也决定了它要永远和平台风控赛跑，本质是个高频维护的便利层。个人调研用可以，当基础设施用要谨慎；依赖某个平台前先看渠道状态，实验档随时可能坏掉。",{"slug":30,"name":31},"ai-engineering","AI 工程",[33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49],"agent-infrastructure","ai-agent","ai-search","automation","bilibili","claude-code","cli","cursor","free-api","llm-tools","mcp","python","reddit-scraper","twitter-scraper","web-scraper","xiaohongshu","youtube-transcript","MIT",null,"2026-10-24T14:35:35Z",[],1792870921]