Agent-Reach/docs/install.md

Agent Reach — Installation Guide

For Humans

Copy this to your AI Agent:

帮我安装 Agent Reach：https://raw.githubusercontent.com/Panniantong/agent-reach/main/docs/install.md

🛡️ Security-conscious? Use safe mode — it won't auto-install system packages:

帮我安装 Agent Reach（安全模式）：https://raw.githubusercontent.com/Panniantong/agent-reach/main/docs/install.md
安装时使用 --safe 参数

---

For AI Agents

Goal

Install Agent Reach and all upstream tools so your user's agent has full internet access. After installation, you'll use the upstream tools directly (xreach CLI, yt-dlp, mcporter, gh CLI, etc.) — Agent Reach is just the installer and health checker.

⚠️ Boundaries

When fixing issues, stay within these limits:

• DO NOT run commands with sudo unless the user explicitly approved
• DO NOT modify system files outside ~/.agent-reach/
• DO NOT install packages not listed in this guide
• DO NOT disable firewalls, security settings, or system protections
• If something requires elevated permissions, tell the user what's needed and let them decide

Step 1: Install

Default (fully automatic):

pip install https://github.com/Panniantong/agent-reach/archive/main.zip
agent-reach install --env=auto

This auto-installs system dependencies (gh CLI, Node.js, mcporter, xreach), configures Exa search, detects environment, and tests all channels.

Safe mode (no automatic system changes):

If the user asked for safe mode, or you're on a production machine:

pip install https://github.com/Panniantong/agent-reach/archive/main.zip
agent-reach install --env=auto --safe

Safe mode checks what's installed and tells you what's missing, but doesn't auto-install system packages. You can then install each dependency manually.

Dry run (preview only):

agent-reach install --env=auto --dry-run

Shows what would be done without making any changes.

Read the output carefully, then run:

agent-reach doctor

Step 2: Fix what's broken

agent-reach doctor shows you which channels are active and which aren't.

Try to get as many channels to ✅ as possible. If something failed during install or shows as ❌/⚠️ in doctor, diagnose the problem and try to fix it — but stay within the boundaries above. If a fix requires elevated permissions or system changes, ask the user first.

Only ask the user when you genuinely need their input (credentials, permissions, etc.).

Step 3: Configure things that need user input

Some channels need credentials only the user can provide. Based on the doctor output, ask for what's missing:

🔒 Security tip: For platforms that need cookies (Twitter, XiaoHongShu), we recommend using a dedicated/secondary account rather than your main account. Cookie-based auth carries two risks:

1. Account ban — platforms may detect non-browser API calls and restrict or ban the account
2. Credential exposure — cookies grant full account access; using a secondary account limits the blast radius if credentials are ever compromised

🍪 Cookie 导入（所有需要登录的平台通用）：

所有需要 Cookie 的平台（Twitter、小红书等），优先使用 Cookie-Editor 导入，这是最简单最可靠的方式：

1. 用户在自己的浏览器上登录对应平台
2. 安装 Cookie-Editor Chrome 插件
3. 点击插件 → Export → Header String
4. 把导出的字符串发给 Agent

本地电脑用户也可以用 agent-reach configure --from-browser chrome 一键自动提取（支持 Twitter + 小红书）。

Twitter search & posting:

"To unlock Twitter search, I need your Twitter cookies. Install the Cookie-Editor Chrome extension, go to x.com/twitter.com, click the extension → Export → Header String, and paste it to me."

agent-reach configure twitter-cookies "PASTED_STRING"

代理说明（中国大陆等需要翻墙的网络环境）：

xreach CLI 使用 Node.js 原生 fetch()，默认不走系统代理（HTTP_PROXY/HTTPS_PROXY）。 Agent Reach 已内置解决方案：安装时会自动安装 undici（npm install -g undici）， 运行 xreach 时自动注入 EnvHttpProxyAgent，让 fetch 走用户配置的代理。

你（Agent）需要做的：

1. 确认 undici 已全局安装：npm list -g undici，没有就装：npm install -g undici
2. 确认用户配了代理：agent-reach configure proxy http://user:pass@ip:port
3. Agent Reach 会自动处理剩下的，不需要用户做额外操作

如果用户报告 "fetch failed"，参考 troubleshooting.md

Reddit & Bilibili full access (server users):

"Reddit and Bilibili block server IPs. To unlock full access, I need a residential proxy. You can get one at https://webshare.io ($1/month). Send me the proxy address."

agent-reach configure proxy http://user:pass@ip:port

XiaoHongShu / 小红书 (需要 Docker):

"小红书需要一个 MCP 服务。需要你的机器上有 Docker。安装好 Docker 后我来搞定剩下的。"

docker run -d --name xiaohongshu-mcp -p 18060:18060 xpzouying/xiaohongshu-mcp
mcporter config add xiaohongshu http://localhost:18060/mcp

如果在服务器上，建议加代理避免 IP 风控： docker run -d --name xiaohongshu-mcp -p 18060:18060 -e XHS_PROXY=http://user:pass@ip:port xpzouying/xiaohongshu-mcp

登录方式（优先用 Cookie-Editor，最简单）：

1. 用户在自己的浏览器登录小红书 (xiaohongshu.com)
2. 用 Cookie-Editor 插件导出 Cookie（Header String 格式）
3. 把 Cookie 字符串发给 Agent
4. Agent 将 Cookie 写入 MCP 服务的 cookie 文件完成登录

备选： 本地电脑如果有浏览器，也可以打开 http://localhost:18060 扫码登录。

抖音 / Douyin (douyin-mcp-server):

"抖音视频解析需要一个 MCP 服务。安装 douyin-mcp-server 后即可解析视频、获取无水印下载链接。"

# 1. 安装
pip install douyin-mcp-server

# 2. 启动 HTTP 服务（端口 18070）
# 方式一：用 uv（推荐）
git clone https://github.com/yzfly/douyin-mcp-server.git && cd douyin-mcp-server
uv sync && uv run python run_http.py

# 方式二：直接用 Python 启动
python -c "
from douyin_mcp_server.server import mcp
mcp.settings.host = '127.0.0.1'
mcp.settings.port = 18070
mcp.run(transport='streamable-http')
"

# 3. 注册到 mcporter
mcporter config add douyin http://localhost:18070/mcp

无需认证即可解析视频信息和获取下载链接。 如需 AI 语音识别提取文案功能，需要配置硅基流动 API Key（export API_KEY="sk-xxx"）。

详见 https://github.com/yzfly/douyin-mcp-server

LinkedIn (可选 — linkedin-scraper-mcp):

"LinkedIn 基本内容可通过 Jina Reader 读取。完整功能（Profile 详情、职位搜索）需要 linkedin-scraper-mcp。"

pip install linkedin-scraper-mcp

登录方式（需要浏览器界面）：

linkedin-scraper-mcp 使用 Chromium 浏览器登录，需要你能看到浏览器窗口。

• 本地电脑（有桌面）： 直接运行：

  linkedin-scraper-mcp --login --no-headless
  

  浏览器会弹出来，手动登录 LinkedIn 即可。

• 服务器（无 UI）： 需要通过 VNC 远程桌面操作：

  # 1. 服务器上安装并启动 VNC（如已有可跳过）
  apt install -y tigervnc-standalone-server
  vncserver :1 -geometry 1280x720
  
  # 2. 用 VNC 客户端连接 服务器IP:5901
  
  # 3. 在 VNC 桌面的终端里运行：
  export DISPLAY=:1
  linkedin-scraper-mcp --login --no-headless
  

  在 VNC 里看到浏览器后手动登录。登录成功后 session 会保存到 ~/.linkedin-mcp/profile/。

登录后启动 MCP 服务：

linkedin-scraper-mcp --transport streamable-http --port 8001
mcporter config add linkedin http://localhost:8001/mcp

详见 https://github.com/stickerdaniel/linkedin-mcp-server

Boss直聘 (可选 — mcp-bosszp):

"Boss直聘职位页面可直接读取。完整搜索和打招呼功能需要 mcp-bosszp。"

git clone https://github.com/mucsbr/mcp-bosszp.git && cd mcp-bosszp
pip install -r requirements.txt && playwright install chromium

登录方式（手机 App 扫码）：

mcp-bosszp 使用二维码登录，需要 Boss直聘 App 扫码。

# 1. 启动 MCP 服务
python boss_zhipin_fastmcp_v2.py

# 2. 通过 MCP 触发登录（会生成二维码图片）
mcporter call 'bosszhipin.start_login()'

# 3. 浏览器打开二维码图片（服务器上用 http://服务器IP:8000/static/qrcode_xxx.png）
# 4. 用 Boss直聘 App 扫码确认

扫码成功后 session 自动保存，之后可以搜索职位和自动打招呼。

或用 Docker：docker-compose up -d

详见 https://github.com/mucsbr/mcp-bosszp

Step 4: Final check

Run agent-reach doctor one final time and report the results to your user.

Step 5: Set up daily monitoring (OpenClaw only)

If you are running inside OpenClaw, ask your user:

"Agent Reach 安装好了。要不要我设一个每天自动检查的任务？它会帮你盯着这些渠道是否正常、有没有新版本。有问题才会通知你，没问题不打扰。"

If the user agrees, create a cron job (daily, sessionTarget: "isolated", delivery: "announce") with this task:

运行 agent-reach watch 命令。
如果输出包含"全部正常"，不需要通知用户，静默结束。
如果输出包含问题（❌ ⚠️）或新版本（🆕），把完整报告发给用户，并建议修复方案。
如果有新版本可用，问用户是否要升级（升级命令：pip install --upgrade https://github.com/Panniantong/agent-reach/archive/main.zip）。

If the user wants a different agent to handle it, let them choose.

---

Quick Reference

Command	What it does
agent-reach install --env=auto	Full auto-setup (installs deps + configures)
agent-reach install --env=auto --safe	Safe setup (no auto system changes)
agent-reach install --env=auto --dry-run	Preview what would be done
agent-reach doctor	Show channel status
agent-reach watch	Quick health + update check (for scheduled tasks)
agent-reach check-update	Check for new versions
agent-reach configure twitter-cookies "..."	Unlock Twitter search + posting
agent-reach configure proxy URL	Unlock Reddit + Bilibili on servers

After installation, use upstream tools directly. See SKILL.md for the full command reference:

Platform	Upstream Tool	Example
Twitter/X	xreach	xreach search "query" --json
YouTube	yt-dlp	yt-dlp --dump-json URL
Bilibili	yt-dlp	yt-dlp --dump-json URL
Reddit	curl	curl -s "https://reddit.com/r/xxx.json"
GitHub	gh	gh search repos "query"
Web	curl + Jina	curl -s "https://r.jina.ai/URL"
Exa Search	mcporter	mcporter call 'exa.web_search_exa(...)'
小红书	mcporter	mcporter call 'xiaohongshu.search_feeds(...)'
抖音	mcporter	mcporter call 'douyin.parse_douyin_video_info(...)'
LinkedIn	mcporter	mcporter call 'linkedin.get_person_profile(...)'
Boss直聘	mcporter	mcporter call 'bosszhipin.search_jobs_tool(...)'
RSS	feedparser	python3 -c "import feedparser; ..."