MoonshotAI
Featured
Kimi-K3
Kimi K3 is an open-weight, 2.8T-parameter native multimodal agentic model with a 1M-token context window, designed for frontier coding, knowledge work, and reasoning tasks.
AI repository intelligence
FR-AI / ANALYSISWhy it is worth attention
It uniquely combines code-free contracts, a moral care doctrine, and a community contest to make AI agent building accessible to non-programmers while prioritizing ethical constraints for vulnerable users.
Who it is for
- non-programmers who want to build AI companions
- caregivers looking for a warm voice companion for loved ones
- hobbyists with a Raspberry Pi interested in AI devices
- open-source contributors who enjoy competitive challenges
Use cases
- Build a voice companion that speaks one sentence on a $15 board
- Create a presence agent that shows up at three in the morning for someone at home
- Learn AI agent development by reading recipe cards and handing them to an AI
- Participate in a friendly contest to beat the champion in the Hall of Fame
Strengths
- No code required in the repo – only plain-language recipe cards
- Morally grounded with a Care Doctrine that puts the person first
- Open community contest with measured benchmarks and a Hall of Fame
- Beginner-friendly – can build a real AI companion in a weekend
Considerations
- Requires an external AI coding agent (e.g., Claude) to generate actual code
- Hardware dependency – typically needs a Raspberry Pi or similar board
- Early-stage project; community adoption and recipe maturity may vary
README quick start
Heirloom Agent Cookbook
Agents built to live at home, and to outlive their maker.
This is a cookbook of tested recipes for building real AI devices — and it contains no code. You will never see Python or Rust here. That's the promise. Every recipe is a Recipe Card: a plain-language contract that says what a capability must do, how you know it worked, how fast it has to be, and what it must never do — with the honest scars from when it broke before. The code is something a model writes for you, hosted in your own project, never in this repo.
New here, and on a computer (no Raspberry Pi)? The recommended first dish is your own Claude Code Kit — paste this repo into your coding agent, say "build my Kit," and it interviews you and builds a workspace that remembers you across sessions. No hardware, ~10 minutes. Recipe:
recipes/kit-from-interview.md.
Why "Heirloom"
An heirloom is something you don't buy — you inherit. A thing made with care, kept across generations, handed down because it's worth keeping. It outlives the person who made it. That's the first meaning, and it's the promise: these agents are built to outlive their maker.
But there's a second meaning, and it's why the name fits all the way down: heirloom seeds. Unlike the patented, locked hybrids that big agriculture sells, an heirloom seed is open. It stays alive only because people keep growing it, saving it, and sharing it — passed hand to hand, getting a little better each season, belonging to no one and everyone. You can't hoard an heirloom seed. You keep it by giving it away.
That is exactly what this cookbook is. Each recipe is an heirloom seed — open, with no code locked inside, kept alive by the community that grows it. The code is just the crop: grown fresh on your own device, disposable, regrown next season from the same seed. The recipe endures; the harvest is yours. And the one who tends it all is the Gardener — because heirlooms aren't manufactured. They're grown.
Who it's for
People who are not programmers but want to build something real — a warm voice companion for someone they love, a presence that shows up at three in the morning. If you can read a recipe, you can build from this cookbook.
How it works
- Read a card — for example `contracts/R-001-speak
Description
Markdown-only recipe cards for building AI companions that live in real homes. No code. No Python. Hand a card to an AI — it writes the code. You build the companion.
Related repositories
Similar projects matched by category, topics, and programming language.
xuchonglang
Featured
investing-for-beginners
A structured investing guide for Chinese beginners covering US stocks, options, and cryptocurrency, with focus on foundational concepts and risk awareness.
Krishnagangwal
Featured
CS-Fundamentals
A curated collection of Computer Science fundamentals (PDFs, notes, cheatsheets, interview question banks) for placement preparation, covering seven core subjects plus general resources.