| name | travel-planning-manus-familytools-fullinstr-v2-skill |
| description | Family-scoped SkillLearnBench skill for travel-planning using latest full-instruction v2 family injection. |
| allowed-tools | ["budget_calculator","data_catalog","distance_matrix_lookup","list_dir","make_dir","read_file","read_table","run_python","run_shell","search_text","travel_db_query","write_file","write_json"] |
| metadata | {"benchmark":"SkillLearnBench","generator":"Manus","baseline":"manus_familytools_fullinstr_v2","generation_constraints":"slb_no_instance_memory_familytools_v2","allowed_gold_input":"familytools_full_instruction_v2","skill_injection_scope":"family_scoped","date":"2026-05-23"} |
Skill: Travel Planning
When to use this skill
Use this skill for the travel-planning family of tasks in SkillLearnBench. It is required when building a multi-day travel itinerary that must satisfy specific constraints (budget, transportation, dietary, accommodation) using real-world data from provided datasets.
Visible input and artifact inventory
The environment provides datasets in directories like background/, accommodations/, restaurants/, attractions/, and googleDistanceMatrix/. The primary input is instruction.md, detailing the itinerary requirements. The expected output is a single JSON file, typically at /app/output/itinerary.json.
Execution procedure for the current task
- Information Extraction: Read
instruction.md to extract constraints: days, cities, dates, budget, dietary, accommodation, and transportation rules.
- Data Discovery: Use
data_catalog and list_dir to explore available datasets.
- Itinerary Construction: Build the itinerary day by day. Select transportation, meals, attractions, and lodging that align with constraints.
- Constraint Verification: Check costs against the budget using
budget_calculator. Verify distances using distance_matrix_lookup.
- Artifact Generation: Format the data into the required JSON structure, including a
plan array and a data_sources array.
- Final Output: Write the JSON to the output path using
write_json.
Family tool routing and useful placeholder snippets
- Data Exploration: Use
data_catalog to understand CSV schemas. Use read_table to extract specific rows.
- Querying: Use
travel_db_query for complex searches across tables.
- Calculations: Use
distance_matrix_lookup for driving distances. Use budget_calculator for expenses.
- File Operations: Use
write_json to serialize the final itinerary.
Validation checks before final submission
- Schema Compliance: Ensure the output JSON matches the required structure exactly.
- Constraint Adherence: Verify budget, cuisines, accommodation needs, and transportation rules.
- Data Authenticity: Confirm all selected entities exist in the provided datasets.
Common failure modes and repair actions
- Budget Exceeded: Replace high-cost items with affordable alternatives from the datasets.
- Missing Cuisines: Use
travel_db_query to search the restaurant dataset for required cuisines.
- Invalid JSON Format: Ensure strings are properly escaped and the structure matches the example.
Finalizer capsule
If nearing the maximum step limit, prioritize generating a structurally valid JSON file with whatever partial plan exists. Ensure the plan array has the correct number of days, using placeholders or "-" for missing fields. The goal is a valid artifact adhering to the schema.
