E-Commerce Last Exam

A benchmark for evaluating LLM agents on 120 real-world travel scene and e-commerce tool-use tasks in isolated Docker environments.

Dataset Code PyPI HF Space FlyAI Platform

Abstract

E-Commerce Last Exam evaluates how well LLM-based agents can solve real-world travel scene and e-commerce tasks using domain-specific CLI tools. Each of the 120 tasks runs inside an isolated Docker container pre-loaded with a SQLite database and purpose-built CLI tools (hotel search, flight booking, product comparison, etc.). Agents must interpret natural-language user requests, call the right tools in the right order, and produce a structured JSON answer scored 0.00–1.00 by an automated verifier. The benchmark covers two domains—travel (77 tasks across hotels, transport, and attractions) and e-commerce (43 tasks across shopping categories)—and is designed to be a reproducible, end-to-end evaluation of agent tool-use capabilities.

120
Total Tasks
77
Travel Tasks
43
E-Commerce Tasks

Evaluation Pipeline

📦
Pull Docker Image
🔧
Sandbox Setup
Permission isolation
🤖
Agent Runs
Tool calls via CLI
✅
Verifier Scores
0.00 – 1.00
📊
Report & Submit

Getting Started

1

Install & Configure

Python ≥ 3.10, Docker required

git clone https://github.com/alibaba-flyai/ecommerce_last_exam.git cd ecommerce_last_exam pip install flyai-bench cp eval_config.yaml eval_config.local.yaml
2

Run Evaluation

Configure your LLM endpoint, then run

flyai-bench --config eval_config.local.yaml run \ --dataset-config travel \ --limit 5
3

Submit Results

Generate report and open a PR

flyai-bench --config eval_config.local.yaml report flyai-bench --config eval_config.local.yaml submit \ --model your-model \ --provider your-provider

Citation

@misc{ecommerce_last_exam, title={E-Commerce Last Exam: A Benchmark for LLM Agent Evaluation on Real-World Tool-Use Tasks}, author={FlyaiLab}, year={2026}, url={https://huggingface.co/datasets/FlyaiLab/ecommerce_last_exam} }