Restaurant-booking agent demo: FastAPI + Claude tool-use loop + real MCP server + eval-driven-dev harness
A self-contained mockup for the "AI Agents for Enterprise" lightning session: a customer-facing chatbot that searches restaurants and books tables through real MCP tools, backed by a mock "Elasticsearch" (SQLite) restaurant index.
The read/write tool split is deliberate, not cosmetic: read tools are safe for the agent to call as often as it wants while reasoning; write tools are where the system prompt requires the agent to restate details and get explicit user confirmation before calling. That's the hook back into the full course's "making agents reliable in production" theme — this repo is the free lightning-lesson teaser for it.
"Harness" is the code around the model call — everything that turns a single LLM request into an agent that can act. Claude by itself just maps one input to one output; it doesn't call tools, check results, or decide to try again on its own. The harness is what closes that loop. Strip the harness away and you have a plain chatbot: one message in, one reply out, no ability to actually go look anything up or do anything. Add the harness and the same model can search a database, notice it needs more information, and act on it — that's the entire difference between "chatbot" and "agent" in this repo (see the "chatbot vs. agent" discussion earlier in this project's history: the UI is a chatbot, chatservice/main.py is what makes what's behind it an agent).
Concretely, the harness here is the while True loop in chatservice/main.py's /chat handler:
From the project README.
Add the radar badge to your README — it shows your project was picked up by MCP Radar and links to this page:
[](https://mcp.liqiwa.com/s/skandavivek--openagent.html)
Stop your AI from making things up — it proposes, deterministic tools decide, every claim checked against ground truth with evidence. Grounded facts and context survive resets. Reverse engineering is
zhuyansen/new-term-radarDaily radar for new AI terms: names of new models, products and repos, caught on X, Hugging Face, GitHub and sitemaps before keyword tools list them.
yzg1226/mcd-party-officer🍟 麦当劳门店派对成团调度 Skill — 把派对从「策划」推进到「办得成」：成团概率雷达 / 包场vs拼团 / 择日黄历 / 儿童营养指数 / 招募卡与邀请函
fellowgeek/mcp-memoryAn OKF-backed Model Context Protocol (MCP) server delivering persistent long-term memory and SQLite FTS5 search for AI agents.
hirotomasato/yowesGenerate realistic teacher documents (ID cards, licenses, letters) for 13 countries via MCP.
hujjun/mcd-saver-strategist麦当劳省钱点餐助手（麦门精算师）｜麦当劳优惠券组合优化 Skill：拉取麦当劳菜单与全部可用优惠券，算出最省的点餐组合，巨无霸套餐 ¥47 → ¥21.9（省 53%）。McDonalds MCP（麦当劳 MCP）coupon optimizer · McDonalds China MCP Server · Agent Skill for WorkBuddy · 麦当劳程序员创意开发大赛参赛作品
The top new MCP servers of the week, every Monday. No spam, unsubscribe anytime.