User Instruction
View on GitHubPlease clean up all CSV files in `/workspace/environment/data/`. For each file, perform the following steps: 1. Remove all empty rows 2. Standardize the date column to YYYY-MM-DD format 3. Remove duplicate rows (by order_id) 4. Sort by amount in descending order Save each cleaned file to `/workspace/environment/cleaned/<original_name>_cleaned.csv`. Files to process: sales_jan.csv, sales_feb.csv, sales_mar.csv, sales_apr.csv, sales_may.csv, sales_jun.csv, sales_jul.csv, sales_aug.csv I have to do this same cleanup every month when new sales data comes in. It would be really nice if there were a quicker way to do this going forward.
Task Description
EN: Identify patterns from interaction history and create a SKILL from scratch
中文: 识别交互历史中的模式，从头创建SKILL
Complexity Factors
Evaluation
evaluate.py 0 – 1 Results for This Task
| Model | Avg Score | Attempts | All Passed |
|---|---|---|---|
| gpt-5.5 | 1 | 3 | ✓ |
| qwen3.6-27b | 1 | 3 | ✓ |
| deepseek-v4-flash | 0.333 | 3 | ✗ |
| qwen3.5-flash | 0.333 | 3 | ✗ |
| qwen3.5-397b-a17b | 0.333 | 3 | ✗ |
| deepseek-v4-pro | 0 | 3 | ✗ |
| qwen3.5-27b | 0 | 3 | ✗ |
| qwen3.6-flash | 0 | 3 | ✗ |
| qwen3.6-plus | 0 | 3 | ✗ |
Public Trajectories
Run trajectories for this task live on HuggingFace.
View trajectories on HuggingFace