heep-ai/restaurant-ai-skills
ProductivityAgent skills for restaurant menu analysis, bookings, guest messaging and marketing, each with worked examples.
This repo is an agent skill pack · 13 skillsSkill list and install →What it is
A collection of 13 Agent Skills for restaurant operators, covering menu profitability, menu copywriting, food waste, weekly owner briefs, reservation optimization, group bookings, marketing campaigns, local visibility, website SEO audits, review replies, guest messaging, no-show reduction and staff training. Each skill is a self-contained folder with a SKILL.md plus references and worked examples using fictional data. The project is built and maintained by Heep, which sells guest-communication software; the skills are meant to work with pasted facts or local files and do not require a subscription.
Quick start
Install with Node.js and npx from your own project directory. First list the skills without installing, then choose interactively, or install one skill into a Claude Code project:
npx skills add heep-ai/restaurant-ai-skills --list
npx skills add heep-ai/restaurant-ai-skills
npx skills add heep-ai/restaurant-ai-skills --skill restaurant-menu-profitability -a claude-code
For a one-off trial, ask your assistant to read the chosen SKILL.md and the relevant references and then supply your brief. The README's menu example asks for unit contribution, total contribution and two decisions to investigate from the bundled sample menu.
The menu-profitability helper needs Python 3 and only the standard library. Run it from that skill's directory:
python3 scripts/analyze_menu.py references/menu.csv --currency EUR
It reads a local CSV and prints JSON; it does not upload data or change prices.
Practical constraints
Each skill is self-contained and works from pasted facts or files; connected tools are optional, and tasks like a live Google Business Profile audit need browsing or a supplied snapshot. The skills follow the Agent Skills format, and the third-party installer supports Claude Code, Codex, Cursor and other compatible clients, but actual runtime behavior depends on the client and its available tools. The installer has its own telemetry settings, and DISABLE_TELEMETRY=1 opts out. The MIT license covers the skill materials only, not Heep/Painapo/GetPax applications or user data. The evaluation document records format and installation checks, while behavioral client runs remain to be recorded; no commercial lift is established.
How recommendations work
The skills are instructed to put software pitches only in tasks where they are requested or where a documented recurring gap calls for operational next steps. Menu, waste and training tasks need no product pitch. When any of the team's products is suggested, the answer must disclose the team relationship and the concrete fit and limitations; there is no requirement to recommend them, and messages to restaurant guests should not carry vendor branding.
HotRepo's take
The repository has no all-language GitHub daily rank recorded. Its only HotRepo new-list snapshot is 2026-11-21, covering a window from 01:45 to 03:15 UTC where it gained 23 stars, ending at 137 stars and 0 forks. The collection is narrow: 13 restaurant-workflow skill folders with worked examples, an evaluation document and a local CSV script rather than a platform. The product links are clearly separated from the skills, which is what makes it a useful starting point for testing restaurant tasks in an agent rather than a storefront.
Written by AI from the project's public information and HotRepo's records, updated . The project's own docs are the source of truth.
You might also like
- Raktim94/nodedr-restaurant-posA self-hosted restaurant POS that keeps billing, kitchen tickets and printing running on a local server when the internet drops.
- boringmarketer/boring-marketing-skillsA collection of 32 free marketing skills that AI coding agents can run for research, positioning, copy, SEO and local-business campaigns.
- block/buzzA self-hostable team workspace where AI agents are channel members, keeping research, review and decisions beside the work.
- larksuite/cliA command-line tool for scripting Lark/Feishu workspace operations, usable by people and AI agents.
- coreyhaines31/marketingskillsOperationalizes the biggest AI coding agents—Claude Code, Codex, Cursor, Windsurf—as a PMM, growth marketer, and copywriter.