Step 1.5: Set Expectations

Before starting any search, output a brief status message so the user knows what to expect:

Researching {city} travel information — this involves multiple searches and will take about 1-2 minutes. Hang tight...

This message must appear before the first search call.

Step 2: Research via Live Search

Use live web search as the primary research method for all research. Target ~45 seconds total for the research phase. Run up to 8 queries max. Prioritize the most impactful queries first.

Use the environment's built-in WebSearch / WebFetch tools or another live search method available to the user.

Do not rely on stale model knowledge when live search is available. Only proceed with partially verified information when a live search method is unavailable or clearly failing, and disclose that in the final confidence section.

Search Strategy — Parallel First

Maximize parallelism to reduce total wait time. Batch searches into 2-3 parallel groups using multiple simultaneous search tool calls:

Batch 1 (simultaneous):

City overview
Weather/climate
Attractions

Batch 2 (simultaneous):

Food/cuisine
Events/festivals
Flight prices (if travel_from provided)

Batch 3 (simultaneous):

Points/miles (if travel_from provided)
Safety advisory

No artificial delays between batches. Only pause if you hit a 429 rate limit (wait 8-15 seconds, retry once).

Query Plan (run in priority order, skip lower-priority if budget exhausted)

"{city}" travel guide overview — city basics, intro
"{city}" travel safety advisory {current_year} — safety, advisories
"{city}" weather climate best time to visit — seasonal info
"{city}" top attractions things to do — sightseeing
"{city}" food must try dishes cuisine — food scene
"{city}" festivals events {current_year} — events calendar
flights from {travel_from} to {city} price — only if travel_from provided
"{city}" points miles award flights from {travel_from} — only if travel_from provided

Source Priority

Prefer these sources (official, authoritative):

travel.state.gov, gov.uk/foreign-travel-advice — travel advisories
Official tourism board sites (visitjapan.jp, etc.)
cdc.gov/travel — health advisories

Secondary (reputable travel content):

lonelyplanet.com, thepointsguy.com, skyscanner.com, google.com/travel
numbeo.com (cost of living), xe.com (currency), rome2rio.com (transit)

Tertiary (use only when primary/secondary lack coverage):

tripadvisor.com

Never use: Reddit, X/Twitter, Facebook, Instagram, TikTok, Quora, Medium, personal blogs

Step 3: Compile Briefing

Write the briefing using all of the following sections in order. If travel_from is NOT provided, skip section 10. If season_month IS provided, tailor sections 3, 6, and 10 to that time window.

Output Format

Use these exact section headings with emojis. Use numbered lists for ranked/ordered content. Use bullet lists for unordered content. Keep paragraphs to 2-3 sentences.

🌍 City Overview

Population, country, language(s), currency, timezone
Brief intro — what the city is known for, its character and vibe
Format population as: 9.7 million or 850,000
Format timezone as: UTC+9 (JST)

📰 Recent History

Notable events from the last ~10 years
Political or economic changes that affect travelers
Major infrastructure changes (new airports, transit lines, etc.)

🗓️ Best Time to Visit

Climate by season with temperature ranges
Peak vs. off-season timing and pricing impact
Weather considerations and natural disaster risks
If season_month provided: focus on that specific window
Format temperatures as: 85°F (29°C) (Fahrenheit first)

🏘️ Top Neighborhoods & Nearby Cities

Notable neighborhoods/districts — 2-3 sentences each covering vibe, key activities, and who it's best for
Each neighborhood/district name must be a named hyperlink to Google Maps
Day-trip cities within 1-2 hours (also with Google Maps links)
Where to stay for different traveler types (budget, luxury, nightlife, culture)

🎯 Things to Do

Top 10-15 attractions, experiences, and landmarks (numbered list)
Mix of iconic must-sees and lesser-known gems
Each attraction name must be a named hyperlink to Google Maps
Include approximate visit duration and cost where known
Format prices in both local currency and USD: ¥1,500 (~$10 USD)

🎉 Popular Events

Major festivals, holidays, and recurring events
If season_month provided: highlight events in that window
Include dates/months when events typically occur
Note which events require advance booking

🍜 Food & Dining

Must-try dishes (numbered list, 8-12 items)
Food culture overview — meal times, dining customs
Price ranges by category: street food, casual, mid-range, fine dining
Tipping norms
Format prices in both local currency and USD

🎌 Cultural Norms

Essential etiquette and customs
Dress codes (temples, restaurants, business)
Do's and don'ts
Communication tips (common phrases, language barriers)
Religious or social sensitivities

🛡️ Safety & Security

General crime overview and safety level
Common scams and tourist traps to watch for
Current travel advisories (cite travel.state.gov or equivalent)
Health considerations (vaccines, water safety, air quality)
Emergency numbers

✈️ Getting There

Only include this section if travel_from is provided.

Direct flight routes and major airlines serving them
Airport info (name, code, distance to city center)
Flight duration
Approximate cash pricing (economy, round trip): $800–$1,200 RT
Points/miles estimates with program names: 60k–80k United MileagePlus miles RT
Transfer partner options: transferable from Chase UR, Amex MR
Best booking strategies and when to book
Airport-to-city transportation options

🚇 Getting Around

Public transit overview (metro, bus, rail) with fare info
Ride-hailing apps available (Uber, local alternatives)
Walkability assessment
Tourist passes or transit cards worth buying
Intercity transportation if relevant

📋 Confidence Notes

Flag data freshness and uncertainty:

Confirmed: Items verified from official/primary sources during this research
Unconfirmed: Items from training data not verified by live search (mark with (unconfirmed))
Conflicting: Items where sources disagreed — note the discrepancy
Stale data flags: Note any data that may change rapidly (prices, exchange rates, political situations)
Include the date of research: Research conducted: {today's date}
Include: Live search queries used: {count}/8
If a fallback live search method was used, name it explicitly

🔗 Sources

List key sources used during research. Use Name — URL format (plain text with full URL). Group by category. Example:

Official: GO TOKYO Official Travel Guide — https://www.gotokyo.org/en/
Official: U.S. State Dept Japan Advisory — https://travel.state.gov/...
Travel guides: Lonely Planet Tokyo — https://www.lonelyplanet.com/...
Points/Miles: The Points Guy — https://thepointsguy.com/...
Flights: Expedia JFK→NRT — https://www.expedia.com/...

Only include sources that were actually consulted. Keep to ~8-12 links max.

Formatting Rules

Emoji section headings: Every H2 uses a relevant emoji prefix
Numbered lists: For ranked/ordered items (top attractions, must-try dishes)
Bullet lists: For unordered items (cultural norms, safety tips)
Bold key terms on first mention: Shinkansen (bullet train)
Italics for foreign words: izakaya (casual bar)
Prices: Always dual currency — ¥1,500 (~$10 USD)
Temperatures: Fahrenheit first — 85°F (29°C)
Distances: Miles first with km — 15 miles (24 km)
Flight durations: 14h 30m
Flight prices: Ranges with RT — $800–$1,200 RT
Points/miles: Program name + amount — 60k–80k United MileagePlus miles RT
Time-sensitive data: Mark with (as of Month YYYY)
Paragraphs: 2-3 sentences max
Google Maps links: For every named location (neighborhoods, attractions, restaurants, airports, stations), use a Markdown named hyperlink on the place name itself: [Sensō-ji Temple](https://www.google.com/maps/search/Sensoji+Temple+Tokyo+Japan). Use + for spaces in the URL. Do NOT put bare URLs at the end of sentences.
No trailing summary: End with Sources section, not a recap

Parameter	Required	Pattern
`city`	Yes	City name (e.g., "Taipei", "Tokyo")
`season_month`	No	"in summer", "in March", "in December"
`travel_from`	No	"from New York", "from JFK", "from LA"

travel-city

/travel-city — City Travel Briefing

Step 1: Parse Inputs