- Home
- Skills
- Agents & Orchestration
- Trip Planner Agent
More screenshots
Works with the AI Agents you already use
Trip Planner Agent
One line in, a source-checked day-by-day trip plan out: a separate critic, emergency card, safety rules, countdown, and HTML/PDF/infographic.
Free
Trip Planner Agent
Example session with this skill installed
Chicago weekend, solo, I am a US citizen --out html
- Read your context and instructions
- Compiled the trip planner agent
- Generated the document
Summary
Real run, unedited, on claude.ai (Opus 5.5, the four skills uploaded under Skills) on 31 Oct 2026. One line in; the skill asked once for the two missing fields (dates, budget), then ran to the end and delivered the HTML plan as a download. The same pack answered a Thai request in Thai the same day (see Case studies in the description).
English example: Chicago weekend (real run, unedited)
Same host, same day (claude.ai, Opus 5.5, 31 Oct 2026). Request: Chicago weekend, solo, I am a US citizen --out html. A chat host has no saved profile, so dates and budget were missing; the skill asked for both in one message, then ran to the end: 13 commands, web searches, validator and critic passed, chicago-weekend-plan.html delivered as a download. Domestic trip, so entry, embassy and plug sections were left out and the omission was explained.
Agent: Two things before I start researching Chicago:
- When? Which weekend? Give exact dates (e.g. "Fri 6 Nov - Sun 8 Nov") or a month like "a weekend in mid-November".
- Budget? An amount in USD, or just low, mid, or high.
Once I have those, I'll build the HTML plan.
Buyer: a weekend in mid-November 2026, mid budget
Agent: Here's the brief:
- Where: Chicago (US)
- When: Fri 13 Nov 2026 -> Sun 15 Nov 2026 (3 days). I picked these dates as the middle weekend of November.
- Who: 1 adult, US passport (from you). Home time zone assumed to be America/Chicago.
- Budget: mid tier, USD
- Style: none given, normal pace. Medicines: none listed, so tell me if you bring any.
Correct me any time; I can revise.
(Searched the web, ran 13 commands, and 5 more tools)
Your Chicago weekend plan is ready, in the file chicago-weekend-plan.html.
Safety: nothing critical came up. The Safety section has two high-priority items:
- Don't use cannabis in public, and don't carry any bought here across state lines or onto your flight.
- Before riding transit alone late at night, check with the CTA or Chicago Police (311) which stations to avoid.
The arrival (11:00 AM Friday at O'Hare) and departure (5:00 PM Sunday) times are examples. Send me your real flights and I'll re-time those two days.
Three highlights:
- Architecture River Cruise, Friday 3:00 PM. The season ends Sun 22 Nov, so you catch one of the last weekends. It finishes right around the 4:30 sunset.
- 360 CHICAGO on Saturday from 3:40. One ticket gets you daylight, sunset and city lights from the 94th floor.
- Food: deep dish at Lou Malnati's on Saturday night, then Italian beef and a Chicago dog at Portillo's on Sunday before the Blue Line home.
Decisions I made for you:
- River North (Freehand Chicago), not the Loop. For a solo traveler it's livelier at night, and you can walk to both dinners and the river. The Loop saves roughly $60 a night and is closer to the airport train, but it's quiet after dark. Say "go with the Loop" to switch.
- A $15 3-Day CTA pass instead of paying per ride. It covers O'Hare both ways, which skips the $5 airport fare, and every ride in between. Paying per ride would cost about $8 more. Say "go with pay per ride" to switch.
What needs your call: book the hotel and the 3:00 PM cruise by Fri 6 Nov. Cruise tickets can't be exchanged, and November departure times weren't published yet. The total comes to about $774 before flights, nearly two-thirds of it the room plus around 19% hotel tax. I didn't judge it against "mid budget" because I had no typical daily figure to compare with.
Next up:
- By Fri 6 Nov: book the hotel and the cruise.
- Tue 10 Nov: re-check opening hours and fares.
- Thu 12 Nov: download the Ventra app and an offline map.
A few more details:
- The theme is autumn maple, for late-autumn Chicago. Say "theme: city" to change it.
- Dates are my pick within mid-November.
- Weather: highs around 50°F, lows around 40°F, about a 1-in-4 chance of a wet day, so I didn't plan rain alternatives.
- Left out because they don't apply to a domestic trip: entry rules, embassy and plug info.
- Not researched: tourist tax-free shopping, travel-health advisories, etiquette.
- No PDF or infographic, since you asked for HTML only.
Say "swap day 2 and 3" or "drop the zoo" and I'll re-plan just that. If you want me to remember your home airport and preferences for next time, run /trip-planner profile.
File delivered: Chicago weekend plan (HTML), Download.
Next steps for the buyer
- Open the HTML, print it or Save as PDF.
- Reply with real flight times; the skill re-times only the arrival and departure days.
- Run
/trip-planner profileonce so the next trip starts with zero questions.
chicago-weekend-plan.html
PDF · document
Example file from a real run - the skill writes it into your workspace.
Connects securely to your tools. The creator never sees your data.
About this skill
What you get
A four-skill pack for a one-time USD 5: trip-planner (the core, which chains the rest automatically), trip-research, trip-critique and trip-render. Give it one line with destination, dates, party, budget and passport ("Lisbon 12–16 March 2027, two adults, both Thai passports, budget 2,000 EUR"), or save those defaults once in a profile, and it runs to the end with no questions. "No questions" means a complete brief: if a companion's passport is unknown it asks once, in one message; a missing budget is asked in that same message, though some hosts assume a mid-range budget instead and state it in the plan. You get a day-by-day itinerary with an emergency card on the first page, traveler-safety dos and don'ts with official sources, a countdown checklist with book-ahead deadlines, and a printable HTML page, PDF and one-page infographic. Say revise afterwards ("swap day 2 and 3") to change it.
Why it is different
Independent tests have reported that chat-generated itineraries often include closed venues, wrong opening hours or impossible routing. This pack refuses to state a travel fact without a fetched source, dates every fact, and runs a separate critic that checks weekday opening hours, day load, connections and deadlines.
- Every choice made for you (area, transport, pace) shows 2-3 options with why, pros, cons and impact, with the pick first.
- Safety content has must-do, must-not-do, legal and hazard items, each with severity and the date it was checked. A rule is stated as fact only with an official source or two independent sources; otherwise it is shown as "verify with <authority>". Dates are spelled out with weekday and month name in the plan's language (Mon 15 Mar 2027 in English), never digits only, and flights show home time.
- Theme follows destination, season or festival, with a readability guard.
- trip-critique also works alone: paste any itinerary and ask it to check for closed days, hours and entry rules.
How it works
- Reads your brief and profile.
- Researches eight topics with cited sources.
- Plans days around real opening hours and travel time.
- A separate critic stress-tests the plan.
- Fixes blockers.
- Builds the emergency card and countdown.
- Renders HTML, PDF and infographic.
Case studies (real runs, 31 Oct 2026)
The demo above is the Claude run, unedited. Here is how the same pack behaved on each host we tried; the gaps are listed as found.
Claude (claude.ai, Opus 5.5, skills uploaded under Skills). Tokyo, 5 days, two Thai passports, Thai prompt. Zero questions, ~10 minutes, 19 commands, validator and critic passed. Safety came first: five critical rules (e-passport, methamphetamine/amphetamine ban, pseudoephedrine over 10 %, codeine over 1 % needs a permit, "verify with MHLW" for sleeping pills, tramadol and CBD), each sourced from the Japanese customs or health pages. Tsukiji was moved off Wednesday, Shinjuku Gyoen off Monday. Two things it said plainly instead of hiding: health topics were not researched this round, and the run hit the host's tool-call cap before attaching files, so one extra "send files" message was needed. Chicago weekend, solo, US passport: asked once for the two missing fields (dates, budget), then delivered the HTML with no entry, embassy or plug sections and said why they were left out, two decisions with switch hints (River North vs the Loop, CTA pass vs pay per ride) and a book-by date for the river cruise. Standalone critique of a pasted 3-day itinerary: three blockers (museum closed that weekday, Ghibli closed Tuesdays and ticket-only, 13:00 departure for a 14:10 flight) and three warnings, all with sources. Known gap: it trusted the weekday written next to the date ("Mon 16 Mar 2027" is really a Tuesday) instead of recomputing it; fixed in 1.0.1.
ChatGPT (Work mode, GPT-6.1 Sol). The same four zips upload at chatgpt.com/skills (Custom GPTs are retired). Same Thai prompt: it stated the 5-line brief with no questions, unzipped the skill, spawned six research subagents (entry, food, health, weather, transport, lodging), moved Tsukiji off Wednesday and reported progress in Thai. The run was cut off at six minutes by the account's 5-hour Work usage limit before rendering, so no files. Plain Chat mode (Instant model) does not activate skills: it answered the Chicago and critique prompts generically, with no sources and no questions, so use Work mode.
Codex CLI (gpt-6.1-sol) and Claude Code, from the release gate the day before: Chicago weekend (domestic) 6/6 checks, Tokyo happy path 15/16, critic on a planted-defect plan 7/7.
Works for everyone
Any nationality (entry and embassy logic follows your passport), your language and your units; dates are spelled out (weekday + month name) in your language. On trips abroad the emergency card is always there: anything that could not be verified is printed as NOT VERIFIED instead of being left out.
Requirements
Any Agent-Skills host, for example Claude Code, claude.ai/Claude Desktop (Skills), ChatGPT (Skills, Work mode), Codex CLI, Gemini CLI or Cursor, with web search and page fetch. Python 3.9+ (standard library only). On Windows, Git Bash or WSL runs everything; a PowerShell-only host runs the Python steps directly and prints the PDF with Chrome or Edge, or tells you how to print it yourself.
Limitations
Install all four skills; the core tells you if one is missing. Plans can be thin when official sites block automated fetching: expect about one verified stop per day and sparse cost estimates in that case; the plan says so in its concerns, and stops whose opening hours could not be verified are marked "Hours not verified — check before you go". Research text (safety rules, notes) may stay in English when a translation into your language is not possible; local names are kept as written locally. The plan is researched from public sources on the date shown; laws and conditions change, so confirm critical items with the embassy or official site before you go. Live facts (hours, prices, entry rules) are dated: re-check them when the Countdown says. It is a planning aid, not legal, medical or immigration advice. PDF/PNG need Chrome, wkhtmltopdf or rsvg-convert, otherwise you get HTML and SVG. Right-to-left languages render best-effort. Chat hosts cap tool calls per turn; a long run may need one "continue" or "send files" message. Screenshots show a sample plan built from demo data.
- Needs web search + page fetch on your host.
- PDF/PNG export uses Chrome/Chromium, wkhtmltopdf or rsvg-convert when installed; otherwise you get HTML (with print styles) and SVG.
- Live facts (hours, prices, entry rules) are fetched at run time and dated; the plan's Countdown tells you what to re-check and when.
- Plans can be thin when official sites block automated fetching (about one verified stop per day, sparse costs); the plan says so in its concerns and marks stops with unverified hours.
- "Zero questions" applies to a complete one-line brief: an unknown companion nationality is asked once; a missing budget is asked, or on some hosts assumed and stated in the plan.
- Research text may stay in English when translation into your language is not possible; local names are kept.
- Dates are spelled out (weekday + month name) in the plan's language, not in a numeric date format you choose.
- Right-to-left languages render best-effort.
- Chat hosts cap tool calls per turn; a long run may need one "continue" or "send files" message.
- ChatGPT: install under chatgpt.com/skills and use Work mode; plain Chat mode does not activate skills.
- Planning aid only, not legal, medical or immigration advice: confirm critical items with the embassy or official site before you go.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Free forever
No account required to browse
Trust & safety
Security scan
Security scan passed 17 days ago
- Free with an Agensi account
Requires