polish: Steve Jobs pass — simplify install to near-zero friction

Key changes:
1. Auto-detect local vs server (SSH/Docker/cloud/headless detection)
   - User no longer needs to answer 'local or server?'
2. Only ask for ONE thing: Exa API key (the highest-value unlock)
3. Everything else is 'ask later' — user says 'set up Twitter' anytime
4. Install guide cut from 200 lines to 100 lines
5. No more overwhelming channel menu — just show what works + one recommendation

Philosophy: default to everything working, only stop when user action is truly needed.
The lazy user path: paste one line → Agent installs → 'want search? get free key at exa.ai' → done.
This commit is contained in:
Panniantong
2026-02-24 06:43:34 +01:00
parent 1b8b202e1e
commit 1d1484fc16
2 changed files with 136 additions and 203 deletions
+63 -16
View File
@@ -70,8 +70,8 @@ def main():
# ── install ── # ── install ──
p_install = sub.add_parser("install", help="One-shot installer with flags") p_install = sub.add_parser("install", help="One-shot installer with flags")
p_install.add_argument("--env", choices=["local", "server"], default="local", p_install.add_argument("--env", choices=["local", "server", "auto"], default="auto",
help="Environment: local computer or server/VPS") help="Environment: local, server, or auto-detect")
p_install.add_argument("--search", choices=["yes", "no"], default="yes", p_install.add_argument("--search", choices=["yes", "no"], default="yes",
help="Enable web search (needs free Exa API key)") help="Enable web search (needs free Exa API key)")
p_install.add_argument("--proxy", default="", p_install.add_argument("--proxy", default="",
@@ -132,6 +132,16 @@ def _cmd_install(args):
print("👁️ Agent Eyes Installer") print("👁️ Agent Eyes Installer")
print("=" * 40) print("=" * 40)
# Auto-detect environment
env = args.env
if env == "auto":
env = _detect_environment()
if env == "server":
print(f"📡 Environment: Server/VPS (auto-detected)")
else:
print(f"💻 Environment: Local computer (auto-detected)")
# Apply flags # Apply flags
if args.exa_key: if args.exa_key:
config.set("exa_api_key", args.exa_key) config.set("exa_api_key", args.exa_key)
@@ -143,16 +153,14 @@ def _cmd_install(args):
print(f"✅ Proxy configured for Reddit + Bilibili") print(f"✅ Proxy configured for Reddit + Bilibili")
# Environment-specific advice # Environment-specific advice
if args.env == "server": if env == "server" and not args.proxy:
print(f"📡 Environment: Server/VPS") print()
if not args.proxy: print("💡 Tip: Reddit and Bilibili block server IPs.")
print(f"⚠️ Reddit and Bilibili block server IPs.") print(" Reddit search still works via Exa (free).")
print(f" To unlock: agent-eyes configure proxy http://user:pass@ip:port") print(" For full access: agent-eyes configure proxy http://user:pass@ip:port")
print(f" Recommend: https://www.webshare.io ($1/month)") print(" Cheap option: https://www.webshare.io ($1/month)")
else:
print(f"💻 Environment: Local computer")
# Test zero-config features # Test channels
print() print()
print("Testing channels...") print("Testing channels...")
results = check_all(config) results = check_all(config)
@@ -161,19 +169,58 @@ def _cmd_install(args):
print(f"{ok}/{total} channels active") print(f"{ok}/{total} channels active")
# What's missing # What's missing
if args.search == "yes" and not args.exa_key: if args.search == "yes" and not args.exa_key and not config.get("exa_api_key"):
print() print()
print("🔍 Search not yet configured. Run:") print("🔍 Recommended: unlock search with a free Exa API key")
print(" agent-eyes configure exa-key YOUR_KEY") print(" agent-eyes configure exa-key YOUR_KEY")
print(" (Get free key: https://exa.ai)") print(" Get free key: https://exa.ai")
# Final status # Final status
print() print()
print(format_report(results)) print(format_report(results))
print() print()
print("✅ Installation complete!") print("✅ Installation complete!")
if ok < total:
print(f" Run `agent-eyes configure` to unlock remaining channels.")
def _detect_environment():
"""Auto-detect if running on local computer or server."""
import os
# Check common server indicators
indicators = 0
# SSH session
if os.environ.get("SSH_CONNECTION") or os.environ.get("SSH_CLIENT"):
indicators += 2
# Docker / container
if os.path.exists("/.dockerenv") or os.path.exists("/run/.containerenv"):
indicators += 2
# No display (headless)
if not os.environ.get("DISPLAY") and not os.environ.get("WAYLAND_DISPLAY"):
indicators += 1
# Cloud VM identifiers
for cloud_file in ["/sys/hypervisor/uuid", "/sys/class/dmi/id/product_name"]:
if os.path.exists(cloud_file):
try:
content = open(cloud_file).read().lower()
if any(x in content for x in ["amazon", "google", "microsoft", "digitalocean", "linode", "vultr", "hetzner"]):
indicators += 2
except:
pass
# systemd-detect-virt
try:
import subprocess
result = subprocess.run(["systemd-detect-virt"], capture_output=True, text=True, timeout=3)
if result.returncode == 0 and result.stdout.strip() != "none":
indicators += 1
except:
pass
return "server" if indicators >= 2 else "local"
def _cmd_configure(args): def _cmd_configure(args):
+73 -187
View File
@@ -2,243 +2,129 @@
## For Humans ## For Humans
Copy and paste this to your AI Agent (Claude Code, OpenClaw, Cursor, etc.): Copy this to your AI Agent:
``` ```
Install and configure Agent Eyes by following the instructions here: Install and configure Agent Eyes: https://raw.githubusercontent.com/Panniantong/agent-eyes/main/docs/install.md
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 ## For AI Agents
You are installing Agent Eyes — giving yourself eyes to see the entire internet. Follow these steps exactly. Minimize questions to the user.
Follow this guide step by step. Do not improvise.
### Step 1: Install ### Step 1: Install + auto-detect
```bash ```bash
pip install git+https://github.com/Panniantong/agent-eyes.git pip install git+https://github.com/Panniantong/agent-eyes.git || pip3 install git+https://github.com/Panniantong/agent-eyes.git
agent-eyes doctor
``` ```
If fails: try `pip3`, or add `--user`, or `--break-system-packages`. Read the doctor output. It tells you:
- What's already working (zero-config channels)
- What needs a key/cookie/proxy
Verify: Also auto-detect the environment:
```bash ```bash
agent-eyes version agent-eyes install --env=auto
``` ```
This command auto-detects local vs server (checks if running in SSH, Docker, or cloud VM).
### Step 2: Detect environment ### Step 2: Show ONE simple summary to user
Ask the user: Based on doctor output, tell the user what's already working and what ONE thing you recommend:
> "First question: are you running on a **local computer** or a **server/VPS**?" > "✅ Agent Eyes installed! You already have: web pages, GitHub, RSS, YouTube.
>
> **One recommendation**: get a free Exa API key to unlock web search + Reddit search + Twitter search.
> → Go to https://exa.ai, sign up (free), and send me the API key.
>
> (I can also set up Twitter, Bilibili, XiaoHongShu, and Reddit later — just ask anytime.)"
This determines what works out of the box and what needs extra setup. **Key principle**: only ask for ONE thing now. Don't overwhelm with choices.
### Step 3: Show channel menu ### Step 3: Configure Exa (if user provides key)
Present this menu to the user. Adjust the details based on their environment (local vs server).
**IMPORTANT**: Make sure the user understands what each tier means and what they get/miss.
---
> **Agent Eyes supports these channels. Let me help you choose what to set up.**
>
> ---
>
> **🟢 Already active — works right now, nothing needed:**
>
> | Channel | What it does |
> |---------|-------------|
> | ✅ Web pages | Read any URL on the internet |
> | ✅ GitHub | Read repos, issues, PRs + search repositories |
> | ✅ RSS | Read any RSS/Atom feed |
>
> These 3 channels are always available. No setup needed.
>
> ---
>
> **🔵 Cookie-based — free, just need to export cookies from your browser (~2 min each):**
>
> | Channel | What it does | Without cookies |
> |---------|-------------|-----------------|
> | 🐦 Twitter/X | Read tweets, search timelines, read threads | Can still read individual tweet URLs (basic mode) |
> | 📺 YouTube | Video transcripts (public + age-restricted) | Public videos work without cookies; age-restricted/member-only videos need cookies |
> | 📺 Bilibili | Video info + subtitles | [LOCAL: works without cookies ✅] [SERVER: ⚠️ blocked — needs proxy, see below] |
> | 📕 XiaoHongShu | Read XHS notes + comments | Falls back to basic web reading (less content) |
>
> **How to export cookies**: open the site in Chrome → F12 → Application → Cookies → copy the values. I'll walk you through it for each one you choose.
>
> ---
>
> **🟡 Free API key — one key unlocks 3 search capabilities (~30 seconds):**
>
> | Channel | What it unlocks |
> |---------|----------------|
> | 🔍 Exa Search (RECOMMENDED) | **Web search + Reddit search + Twitter search** — one free key powers all three |
>
> Without Exa: I can read URLs you give me, but I cannot search the internet for you.
> With Exa: I can find information across the entire web, Reddit, and Twitter.
>
> Free tier: 1000 searches/month at https://exa.ai — more than enough for most users.
>
> ---
>
> **🟠 Proxy-based — needed only on servers, $1/month shared across channels:**
>
> | Channel | Why proxy needed | Without proxy |
> |---------|-----------------|---------------|
> | 📖 Reddit | Reddit blocks server IPs | **Reddit search still works via Exa (free!)** — you just can't read full posts with all comments |
> | 📺 Bilibili | Bilibili blocks server IPs | Won't work from server |
>
> [ONLY SHOW THIS SECTION IF USER SAID "SERVER"]
> [IF LOCAL: Skip this section — Reddit and Bilibili work directly on local computers]
>
> One proxy works for both. Recommend: https://www.webshare.io ($1/month residential proxy).
> Same proxy also works for Bilibili cookies-based access.
>
> ---
>
> **Which channels do you want me to set up?**
> (I recommend: **Exa Search** for sure, plus any cookie-based channels you actually use)
---
Wait for the user to choose. Then go to Step 4.
### Step 4: Run base installer
```bash
agent-eyes install --env=<local|server>
```
### Step 5: Configure each selected channel
Run the corresponding command for each channel the user chose.
Every configure command auto-tests and gives instant feedback (✅ or ❌).
---
#### Exa Search
Tell user: "Go to https://exa.ai, sign up (free), copy your API Key, and send it to me."
```bash ```bash
agent-eyes configure exa-key THE_KEY agent-eyes configure exa-key THE_KEY
``` ```
--- After success, tell user:
#### Twitter/X > "✅ Done! I can now search the web, Reddit, and Twitter for you.
>
> Want me to set up any of these too? (just say which ones)
> - 🐦 Twitter deep search (need browser cookies, 2 min)
> - 📕 XiaoHongShu (need browser cookies, 2 min)
> - 📺 Bilibili (need browser cookies, 2 min)
> - 📖 Reddit full reader (need $1/month proxy — search already works via Exa!)
>
> Or just say 'all' and I'll walk you through each one."
Tell user: ### Step 4: Configure additional channels (only if user asks)
> "Open https://x.com in Chrome (make sure you're logged in).
> Press F12 → Application → Cookies → https://x.com
> Find these two values: **auth_token** and **ct0**
> Copy both and send them to me."
For each channel the user asks for, use the appropriate command:
**Twitter:**
Ask: "Open x.com in Chrome → F12 → Application → Cookies → copy **auth_token** and **ct0**"
```bash ```bash
agent-eyes configure twitter-cookies AUTH_TOKEN CT0 agent-eyes configure twitter-cookies AUTH_TOKEN CT0
``` ```
--- **XiaoHongShu:**
Ask: "Open xiaohongshu.com in Chrome → F12 → Network → click any request → copy the Cookie header value"
#### YouTube (cookies for age-restricted videos) ```bash
agent-eyes configure xhs-cookie COOKIE_STRING
Tell user: ```
> "Most YouTube videos work without any setup. If you want access to age-restricted or member-only videos:
> Which browser do you use? (chrome/firefox/edge/safari)"
**YouTube (age-restricted only):**
Ask: "Which browser do you use?"
```bash ```bash
agent-eyes configure youtube-cookies chrome agent-eyes configure youtube-cookies chrome
``` ```
(replace `chrome` with whatever browser they use — yt-dlp reads cookies directly from the browser)
--- **Bilibili:**
- Auto-detected as local → already works, nothing needed
#### Bilibili - Auto-detected as server → needs proxy (same as Reddit, see below)
**If local**: Already works. No setup needed.
**If server**: Needs proxy (same proxy as Reddit, see Proxy section below).
If user also wants cookies for member-only content:
> "Open https://bilibili.com in Chrome (logged in).
> F12 → Application → Cookies → find **SESSDATA**
> Send it to me."
(For now, Bilibili public API works without cookies. Cookie support is for future member-only content.)
---
#### XiaoHongShu
Tell user:
> "Open https://www.xiaohongshu.com in Chrome (make sure you're logged in).
> Press F12 → Application → Cookies → https://www.xiaohongshu.com
> Select all cookies, right-click → Copy All
> Or just copy the full cookie string from a network request header.
> Send it to me."
**Reddit full reader + Bilibili (server only):**
Ask: "Reddit/Bilibili block server IPs. Get a $1/month residential proxy at webshare.io, then send me the proxy address."
```bash ```bash
agent-eyes configure xhs-cookie THE_COOKIE_STRING agent-eyes configure proxy http://user:pass@ip:port
``` ```
--- ### Step 5: Final status
#### Proxy (Reddit + Bilibili on server)
Tell user:
> "Reddit and Bilibili block server IPs. You need a residential proxy to access them.
> 1. Go to https://www.webshare.io
> 2. Sign up and get a residential proxy ($1/month is enough)
> 3. Copy your proxy address (format: http://username:password@ip:port)
> 4. Send it to me"
```bash
agent-eyes configure proxy THE_PROXY_URL
```
This configures the proxy for both Reddit and Bilibili at once.
---
### Step 6: Final health check
```bash ```bash
agent-eyes doctor agent-eyes doctor
``` ```
Summarize what's working for the user. Example: > "✅ All set! Here's what I can do:
> [list active channels]
> "✅ All done! Here's what I can do now:
> >
> 📖 **Read**: Web, GitHub, YouTube, Bilibili, Twitter, XiaoHongShu, RSS > Try: 'search Reddit for ...' or 'read this URL: ...'"
> 🔍 **Search**: Web, GitHub, Reddit, Twitter
>
> Try it out — ask me to search or read anything!"
--- ---
## Configure anything later
Users can ask their Agent to configure new channels anytime:
- "Help me set up Twitter for Agent Eyes"
- "I got a proxy, configure it for Agent Eyes"
- "Add my XiaoHongShu cookies"
The Agent runs `agent-eyes configure <channel> <value>` — each command auto-tests.
## Quick Reference ## Quick Reference
| Action | Command | | Command | What it does |
|--------|---------| |---------|-------------|
| Install | `pip install git+https://github.com/Panniantong/agent-eyes.git` | | `agent-eyes doctor` | Show what's working |
| Base setup | `agent-eyes install --env=<local\|server>` | | `agent-eyes configure exa-key KEY` | Unlock web/Reddit/Twitter search |
| Add Exa key | `agent-eyes configure exa-key YOUR_KEY` | | `agent-eyes configure twitter-cookies TOKEN CT0` | Unlock Twitter deep search |
| Add Twitter | `agent-eyes configure twitter-cookies AUTH_TOKEN CT0` | | `agent-eyes configure xhs-cookie COOKIE` | Unlock XiaoHongShu |
| Add YouTube | `agent-eyes configure youtube-cookies chrome` | | `agent-eyes configure youtube-cookies chrome` | Unlock age-restricted YouTube |
| Add XiaoHongShu | `agent-eyes configure xhs-cookie COOKIE_STRING` | | `agent-eyes configure proxy URL` | Unlock Reddit + Bilibili (server) |
| Add proxy | `agent-eyes configure proxy http://user:pass@ip:port` | | `agent-eyes read <url>` | Read any URL |
| Health check | `agent-eyes doctor` | | `agent-eyes search "query"` | Search the web |
| Read URL | `agent-eyes read <url>` |
| 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"` |