What problem does it solve? Planning a trip requires verifying dozens of volatile facts — restaurant operating status, hotel rates, transport schedules, visa rules, attraction booking windows — and AI assistants often fabricate these details from training data. This Skill enforces a strict evidence standard so every price, schedule, and venue in a travel plan is traceable to a real source with a research date, or explicitly degraded to a labeled fallback. ## Core Features & Use Cases - Three output modes: planning-only advice, full guide-redesign (markdown + HTML deliverables), and existing-page refactor that preserves all user-supplied content while reorganizing it. - Domain rule packs: dedicated references for dining verification, hotel tiering with hardware-freshness signals, round-trip transport, budget category splits with FX timing, attractions booking lead times, local specialties, safety and emergency info, and trip prep (visa, payment, SIM, insurance, etiquette). - Anti-fabrication pipeline: knowledge-layer classification, parallel batch verification sliced by geography, login-wall channel ladders, timeout degradation to search advisory cards, and a mechanical HTML validator plus a final-check assertion list. - Use Case: Ask it to plan a 5-day family trip to Tokyo over Spring Festival with a budget — it produces a day-by-day guide with verified restaurants, hotel shortlists, transport cards, a pre-trip recheck block, and a full safety section, each fact cited with source and date. ## Quick Start Ask the assistant to plan a trip with your destination, dates, party, and budget, for example: plan a 5-day Tokyo trip for two over Golden Week with a 50000 CNY budget and a focus on dining.
