# Agent Eyes — Installation Guide ## For Humans Copy and paste this to your AI Agent (Claude Code, OpenClaw, Cursor, etc.): ``` Install and configure Agent Eyes by following the instructions here: https://raw.githubusercontent.com/Panniantong/agent-eyes/main/docs/install.md ``` Your Agent will handle everything. You just answer a few questions. --- ## For AI Agents You are installing Agent Eyes — giving yourself eyes to see the entire internet. Follow this guide step by step. Do not skip steps. ### Step 1: Install ```bash pip install git+https://github.com/Panniantong/agent-eyes.git ``` If fails: try `pip3`, or add `--user`, or `--break-system-packages`. Verify: ```bash agent-eyes version ``` ### Step 2: Ask environment Ask the user: > "You're installing Agent Eyes! First question: are you running on a **local computer** or a **server/VPS**?" Remember their answer — it affects which channels need extra setup. ### Step 3: Show channel menu Show the user this menu. Explain each option clearly. **Present it like this** (adjust the Bilibili line based on Step 2 answer): --- > Here are all the channels Agent Eyes supports. I'll help you set up whichever ones you want. > > **Already active — no setup needed ✅** > These work right now, nothing to configure: > - ✅ **Web pages** — read any URL > - ✅ **GitHub** — read repos, issues, PRs + search repos > - ✅ **RSS** — read any RSS/Atom feed > - ✅ **YouTube** — video transcripts via yt-dlp > - ✅ **Twitter basic** — read individual tweets > > **Recommended — free, takes 30 seconds each 🔍** > - 🔍 **Exa Search** (STRONGLY RECOMMENDED) > - What it unlocks: **full web search + Reddit search + Twitter search** > - What you need: one free API key from https://exa.ai (1000 searches/month free) > - Without it: I can read URLs you give me, but can't search the internet for you > > **Optional — easy setup 🔧** > - 🐦 **Twitter Advanced** — search timelines, read threads, deep search > - What you need: export cookies from your browser (free, 2 minutes) > - Without it: I can still search Twitter via Exa and read individual tweets > > - 📺 **Bilibili** [IF LOCAL: "already works ✅"] [IF SERVER: "⚠️ needs proxy — Bilibili blocks server IPs. Proxy costs ~$1/month at webshare.io. Without it, Bilibili won't work from your server."] > > **Optional — more setup 🔨** > - 📖 **Reddit Full Reader** — read complete posts + all comments > - What you need: residential proxy (~$1/month at https://webshare.io) > - Without it: I can still **search** Reddit content via Exa (free!), just can't read full posts with all comments > - [IF SERVER: "Same proxy works for both Reddit and Bilibili"] > > - 💬 **WeChat** — read WeChat public articles > - What you need: install browser component (~150MB download) > > - 📕 **XiaoHongShu** — read XHS notes > - What you need: browser component + scan QR code once to login > > **Which ones would you like me to set up?** (I recommend at least Exa Search) --- Wait for user to choose. Then proceed to Step 4. ### Step 4: Run installer ```bash agent-eyes install --env= ``` ### Step 5: Configure selected channels For each channel the user selected, run the corresponding configure command. Each command auto-tests after configuring. #### If user selected: Exa Search Tell the user: > "Go to https://exa.ai, sign up (free), and send me your API Key." When they provide it: ```bash agent-eyes configure exa-key THE_KEY ``` Expected: `✅ exa-key configured! Testing search... ✅ Search works!` #### If user selected: Reddit Full Reader and/or Bilibili (on server) Tell the user: > "Reddit and Bilibili need a residential proxy. Steps: > 1. Go to https://www.webshare.io > 2. Sign up and get a proxy ($1/month plan is enough) > 3. Copy your proxy address (format: http://username:password@ip:port) > 4. Send it to me" When they provide it: ```bash agent-eyes configure proxy THE_PROXY_URL ``` Expected: `✅ proxy configured! Testing Reddit... ✅ Reddit accessible!` #### If user selected: Twitter Advanced Tell the user: > "Twitter Advanced needs cookies from your browser. Steps: > 1. Open https://x.com in Chrome and make sure you're logged in > 2. Press F12 → Application tab → Cookies → https://x.com > 3. Find and copy these two values: **auth_token** and **ct0** > 4. Send them to me" When they provide cookies: ```bash # birdx cookie setup (if birdx is available) which birdx || pip install birdx # Configure birdx with cookies (method varies by birdx version) ``` #### If user selected: WeChat ```bash pip install agent-eyes[browser] playwright install chromium ``` Tell user: "✅ WeChat is ready! Send me any WeChat article URL." #### If user selected: XiaoHongShu ```bash pip install agent-eyes[browser] playwright install chromium ``` Tell user: "I'll need you to scan a QR code once to log in. Ready?" Then read the guide: read the file at `agent_eyes/guides/setup-xiaohongshu.md` for detailed steps. ### Step 6: Final check ```bash agent-eyes doctor ``` Summarize results for the user naturally. Example: > "✅ All done! Here's what I can do now: > - 📖 Read: Web, GitHub, YouTube, RSS, Tweets, Reddit, Bilibili > - 🔍 Search: Web, GitHub, Reddit, Twitter > > Try asking me things like: > - 'Search Reddit for best self-hosted LLM' > - 'Read this GitHub repo: [URL]' > - 'What's trending on Twitter about AI agents?'" --- ## Quick Reference | Action | Command | |--------|---------| | Install | `pip install git+https://github.com/Panniantong/agent-eyes.git` | | Setup | `agent-eyes install --env=` | | Add Exa key | `agent-eyes configure exa-key YOUR_KEY` | | Add proxy | `agent-eyes configure proxy http://user:pass@ip:port` | | Health check | `agent-eyes doctor` | | Read any URL | `agent-eyes read ` | | Search web | `agent-eyes search "query"` | | Search GitHub | `agent-eyes search-github "query"` | | Search Reddit | `agent-eyes search-reddit "query"` | | Search Twitter | `agent-eyes search-twitter "query"` |