Travel planning workflow

Skill cxcscmu/SkillLearnBench/skills/human_authored/travel-planning/travel-planning-workflow

[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 travel-planning-workflow

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

Invoke this skill when performing travel planning. It defines a structured workflow for generating itineraries. Following this workflow is essential to ensure the final plan satisfies user requirements.

SKILL.md

2.9 KB, 527 tokens by cl100k_base, as published. Nobody here has run it

Overview

This skill provides a travel planning workflow. It is important to follow this workflow. IMPORTANT:Do not forget to perform final validation and refinement.

Travel Planning Workflow

Step 1: User Preference and Constraint Analysis

Extract the user's requirements and implicit preferences, including: Trip duration, Number of travelers, Budget constraints, Transportation constraints (e.g., no flights), Special requirements (e.g., pet-friendly accommodations), Food preferences (e.g., American, Mediterranean, Chinese, Italian)

Default assumption: If not explicitly specified, each day should include at least one activity (i.e., no empty days).

Step 2: Global Planning and Sequential Scheduling

2.1 Global Skeleton Construction

Select candidate cities based on user constraints and dataset availability
Construct a preliminary travel route (considering distance and feasibility)
Allocate the total number of days across selected cities as a draft plan

2.2 Sequential Day-by-Day Planning

Generate the itinerary in chronological order:

Using the provided tools, and for each day:

Assign one attraction per day, unless the user explicitly indicates that they do not want to go out.
Assign meals based on cuisine preferences
Assign accommodation (must satisfy constraints such as pet-friendly)

2.3 Iterative Adjustment and Backtracking

During planning, continuously validate feasibility. For example:

If a city does not have enough attractions → reduce its allocated days
If all cities combined still lack sufficient attractions → reselect cities
If constraints are violated (budget, accomodations, etc.) → adjust previous steps

This step introduces a feedback loop, allowing the system to revise earlier decisions.

Step 3: Important!! Final Validation and Refinement

Before returning the final itinerary, the agent must perform an explicit final validation pass. The validation should check:

All user constraints must be satisfied, including transportation constraints (e.g., no flights), special requirements (e.g., pet-friendly accommodations, which should be explicitly reflected using keywords such as pet or animal), and food preferences (e.g., American, Mediterranean, Chinese, and Italian cuisines).
Each day includes at least one attraction, including both the first day and last day (unless the user explicitly indicates that they do not want to go out.)
Travel transitions between cities are feasible
The itinerary format is complete and consistent, with no missing or contradictory information.

Please refine the itinerary if any violations or inconsistencies are detected