[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"$f15n51ym7sp9x2":3},{"code":4,"message":5,"data":6,"timestamp":26},10000,"success",{"full_name":7,"position":8,"description":9,"language":10,"stars":11,"forks":12,"stars_gained":13,"age_days":14,"hn_threads":15,"one_liner":16,"summary_md":17,"topic":18,"github_topics":21,"license":22,"homepage":23,"updated_at":24,"history":25},"zcbacxc\u002Fmcd-party-planner",16,"麦当劳 MCP 聚会\u002F团餐组织助手 — 基于 Model Context Protocol 的派对预订与团餐规划 CLI\u002FSkill","Python",1499,70,158,9,[],"A CLI\u002Fskill that plans McDonald's group meals and party bookings using real MCP pricing for organizers.","## The problem\nOrganizing a group meal at a fast-food chain means juggling three things at once: which nearby branch is open, what combination of items feeds N people, and whether the total stays under budget. Doing this by hand across an app, a menu and a calculator is slow and error-prone. This project takes city, keyword, headcount and budget, queries the McDonald's MCP for nearby stores and live menus, and returns several priced meal plans with per-person cost and an over\u002Funder-budget flag.\n\n## What's different\nMost \"agent skill\" repos are prompt wrappers whose numbers are invented or hard-coded. Here every price comes from the service's own `calculate-price` tool, and items blocked by purchase limits are skipped with a message rather than estimated. It also chains five separate queries to resolve a party booking (product → city → store → date → session) and refuses to submit orders unless explicitly confirmed, with no retry on 5xx for writes. The calling layer adds throttling and 429 backoff. That discipline is the interesting part, not the feature list.\n\n## Who it's for\nUseful if you are the person who runs team lunches, class outings or a kid's birthday party at McDonald's in China and want comparable, real quotes in seconds. It also works as a reference for MCP developers who want to see a skill orchestrate roughly fifteen tools with proper rate limiting and dry-run defaults. Look elsewhere if you are outside mainland China, need multi-vendor catering, or want a polished UI — this is a CLI and a WorkBuddy skill, not an app.\n\n## Cost of adoption\nYou need a free token from open.mcd.cn before anything works; without it only unit tests run. Python 3.9+, standard library only, no pip install. The README is unusually honest that CI green means offline unit tests only, and that the real smoke test must be run locally with a token. It is a single-developer contest entry, MIT licensed, with no stated maintenance commitment, and it depends entirely on a third-party MCP whose menus and prices rotate daily, so output will not reproduce.\n\n## Verdict\nThis is a nine-day-old contest entry that picked up about 1,400 stars, with a momentum signal of roughly 153 stars per day at its peak — a spike, not an established tool. The engineering hygiene is above average for a hackathon project: dry-run defaults, honest CI caveats, no fabricated prices. But it is a thin layer over someone else's API, so its lifespan is tied to that API and to the author's interest after the contest. Worth reading for the MCP orchestration pattern; adopt it only if you actually organize McDonald's group meals in China and accept that it may go stale.",{"slug":19,"name":20},"ai-engineering","AI Engineering",[],"MIT",null,"2026-10-25T02:05:23Z",[],1792922132]