Run2 itinerary validation check

[COLM'26] SkillLearnBench is the first benchmark for evaluating continual learning methods that automatically generate agent skills.

Install

npx -y skills add cxcscmu/SkillLearnBench --skill run2_itinerary-validation-check

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

What its author says it does

Copied from the file, not written here

A final verification step to ensure the JSON output meets all structural and constraint requirements before delivery.

1.1 KB, 233 tokens by cl100k_base, as published. Nobody here has run it

Day Count Verification: Ensure the plan array contains exactly 7 objects representing Day 1 to Day 7.
Attraction Field Validation:
- Iterate through every day object.
- Verify that the attraction string is not empty ("").
- Verify that the attraction string is not a null-placeholder ("-").
- Confirm the trailing semicolon format.
Constraint Check:
- Confirm transportation never mentions "flight" or "plane".
- Confirm accommodation is sourced from the database and is pet-friendly.
- Confirm data_sources lists all files accessed (e.g., distance matrix, restaurant csv, etc.).
City Consistency: Ensure that if the current_city is "from A to B", the restaurants and attractions for that day are located in A, B, or a city on the path between them.
Budget Check: Sum the estimated costs of restaurants and accommodations (if price data is available in the CSV) to ensure the total is within the $5,100 limit.