Lt Shop Support — AI skill for Claude Code

Super Fit Support Agent

A customer support agent for the LT Shop men's apparel store (https://shop.lifetime.life/apparel-accessories/men-s-apparel), built with the [Claude Agent SDK](https://code.claude.com/docs/en/agent-sdk/python). It answers questions strictly from `qa_knowledge_base.json` and escalates anything else to a human.

It ships in two forms that share the same knowledge base, system prompt and tools:

• main.py — a terminal chat on the Agent SDK (the reference implementation).
• web/ — a Next.js chat UI that deploys to Vercel. Live demo: https://super-fit-coral.vercel.app

This is a concept project by Roy Torres, not affiliated with Life Time, Inc. The web UI shares its design system with the [LT Shop concept](https://roy-torres.github.io/lt-shop-concept/) ([source](https://github.com/roy-torres/lt-shop-concept)); see [Design](#design) below.

Project layout

super-fit/
├── main.py                 # Agent SDK agent + terminal chat loop
├── server.py               # HTTP/SSE front for main.py (optional, for the web app)
├── qa_knowledge_base.json  # business info + FAQ entries (single source of truth)
├── escalations.jsonl       # tickets written by escalate_to_human (gitignored)
├── requirements.txt
├── .env.example            # ANTHROPIC_API_KEY
└── web/                    # Next.js chat UI (see "Web app" below)

Prerequisites

• Python 3.10+
• Claude Code CLI — the SDK runs on top of it: npm install -g @anthropic-ai/claude-code
• An Anthropic API key (or an existing Claude Code login)
• Node.js 20+ — only for the web app

Setup

python3 -m venv venv
source venv/bin/activate
pip install -r requirements.txt
cp .env.example .env   # then paste your key from https://console.anthropic.com/

Run

Terminal chat:

python main.py

Web app (see [Web app](#web-app) below):

cd web && npm install && cp .env.example .env.local && npm run dev

Example:

You: ho