VIVPM/ecommerce-agent
An AI-powered e-commerce agent that can handle complex user queries about products, inventory, and order status. Built with a FastAPI backend and a React (Vite) frontend, featuring integration with LLMs for natural language understanding and automated task execution.
What's novel
An AI-powered e-commerce agent that can handle complex user queries about products, inventory, and order status. Built with a FastAPI backend and a React (Vite) frontend, featuring integration with LLMs for natural language understanding and automated task execution.
Code Analysis
10 files read · 3 roundsAn AI-powered e-commerce chat assistant that routes user queries via LLM to one of three tools (text-to-SQL product search, RAG-based FAQ, or saved-product comparison), with streaming responses, context memory, and a React frontend.
Strengths
Genuinely sophisticated prompt engineering (Bayesian confidence weighting for ratings, careful NULL handling, prevention of false matches) and a clean, well-documented architecture with provider abstraction, Postgres-backed caching, and fail-open observability.
Weaknesses
Test coverage is thin — only a logging test and an evaluation harness exist, with no unit tests for the core routing, SQL generation, or memory rewriting logic; the project is also tightly coupled to Gemini/Pinecone despite the provider abstraction.
Score Breakdown
Signal breakdown
Innovation
Craft
Traction
Scope
Evidence
Commits
28
Contributors
1
Files
42
Active weeks
2
Repository
Language
Python
Stars
1
Forks
0
License
MIT