Fliggy travel search, wired straight into your coding agent

What it does

FlyAI is a skill package that connects Claude Code, OpenClaw, and other skill-compatible agents to Fliggy, Alibaba’s travel platform. You ask in natural language — English or Chinese — and get structured JSON with real-time inventory: flights, trains, hotels, attractions, cruises, even visas and SIM cards. Every result carries a direct booking link, so the agent hands you something actionable, not a travel brochure.

The interesting bit

It’s a genuinely practical use of the agent-skills pattern: instead of the agent scraping a website or hallucinating prices, it calls one of eight purpose-built commands backed by real Fliggy inventory. ai-search handles messy multi-constraint intent (“budget 2000 per person, near West Lake”), while the structured commands like search-flight expose deep filtering — cabin class, layover city, hour ranges, price ceilings, eight sort modes.

Key highlights

• Eight commands: broad keyword-search, semantic ai-search, plus structured flight, train, hotel, POI, and Marriott-specific searches
• Works with no API key; an optional one unlocks enhanced results
• Bilingual queries — plain English or Chinese both work
• Flight and train searches support fine-grained filters: departure-hour windows, max duration, price caps, direct-only
• Hotel search filters by star rating, bed type, nearby attraction, and price

Caveats

• The inventory is Fliggy’s, so it’s strongest for China-centric travel; prices are in CNY
• Some filter values (hotel types, POI categories) are Chinese-language enums, which may trip up purely English workflows
• It’s glue code by design — the value is the Fliggy API access, not novel engineering

Verdict

Worth a look if you live in your terminal and book China-region travel. If your agent stack doesn’t support skills, or Fliggy isn’t in your orbit, there’s little here for you.