clawhub install scientific-internet-accessInstall this skill, and your OpenClaw AI agent becomes a network access specialist.
Users say "上不了Google" or "帮我翻墙" — your agent handles the entire flow:
Step 1: Ask device type (or detect from screenshot)
Step 2: Recommend client app with exact download link
Step 3: Auto-fetch fastest nodes (scraped + tested live)
Step 4: Guide import — one tap at a time
- ONE step at a time. Never dump 5 instructions. One action, wait for confirmation, next.
- Every instruction includes HOW. Not "截图发我" but "同时按住电源键+音量上键，屏幕闪一下就截好了，然后发给我."
- Stuck = downgrade, never repeat. User says 不懂 → rephrase simpler → teach screenshot → teach voice.
The skill bundles four Python scripts that work together:
| Script | Purpose |
|---|---|
scraper.py |
Monitors 10+ public node subscription sources, fetches and decodes nodes |
tester.py |
Runs 20 parallel TCP connectivity tests, ranks by latency |
formatter.py |
Outputs in 6 formats: Clash, V2Ray, Surge, Shadowrocket, Base64, plain text |
handler.py |
Orchestrates the flow: scrape → test → format → present to user |
All network connections are to public node subscription URLs (GitHub raw content, public APIs). The tester verifies connectivity to the parsed node endpoints. No credentials are required or stored.
- No environment variables or credentials required
- All source URLs are public and auditable in
scraper.py - Node data is stored locally in
~/.openclaw/workspace/ - No system-wide configuration changes
- Review the scripts yourself before installing: they're short and readable
| Metric | Value |
|---|---|
| Node sources monitored | 10+ |
| Parallel test threads | 20 |
| Output formats | 6 |
| Supported protocols | VMess, VLESS, Trojan, Shadowsocks |
| Scrape interval | Every 2 hours |
| Health check | Every 30 minutes |
| User setup steps | 3 |
| Cost | $0 |
~/.openclaw/skills/scientific-internet-access/
├── SKILL.md # Agent instructions
├── scripts/
│ ├── handler.py # Interaction orchestrator
│ ├── scraper.py # Multi-source node scraper
│ ├── tester.py # Parallel connectivity tester
│ └── formatter.py # Multi-format output generator
└── .github/workflows/
├── publish.yml # Tag → Release → ClawHub
└── sync-magic.yml # Mirror sync
| Protocol | iOS | Android | Windows | macOS |
|---|---|---|---|---|
| VMess | Shadowrocket | v2rayNG | v2rayN / Clash Verge | ClashX / V2rayU |
| VLESS | Shadowrocket | v2rayNG | v2rayN | V2rayU |
| Trojan | Shadowrocket | v2rayNG | v2rayN | ClashX |
| Shadowsocks | Shadowrocket | v2rayNG | Clash Verge | ClashX |
Magic Internet Access — same engine, auto-synced via GitHub Actions on every tag.
Claude follows the step-by-step instructions more reliably. Gemini tends to improvise.
clawhub install scientific-internet-accessOr manually:
git clone https://github.com/shadowrocketai/Scientific-Internet-Access.git ~/.openclaw/skills/scientific-internet-accessMIT