Skip to content

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

89 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

name solo-unicorn-builder
description Free, open-source AI command center for builders who've never shipped an impactful service end-to-end. 36 core AI-powered skills plus 7 add-on skills spanning engineering, product, marketing, sales, finance, and operations — AI fills the gaps between your specialty and everything else it takes to go from requirement to production.

Solo Unicorn Builder

You're great at your specialty. But you've never shipped an impactful service end-to-end — because someone else always handled the parts outside your lane.

Solo Unicorn Builder is a free, open-source command center that gives you 36 core AI-powered skills plus 7 add-on skills — 43 total — spanning engineering, product, marketing, sales, finance, legal, and operations. AI fills the gaps between your expertise and everything else it takes to go from requirement to production.

Who This Is For

Engineers, technical specialists, product managers, business process owners, entrepreneurs, and solution builders who can do their job — but have never carried an impactful service from requirement to production on their own. You're a frontend engineer who's never set up a CI/CD pipeline. A backend developer who's never written a PRD. A DevOps specialist who's never validated a product idea. A product manager who's never deployed to the cloud. A business owner who's never turned a workflow into software.

You're good at what you do. But the work outside your specialty — requirements gathering, design, testing, deployment, marketing, legal, finance — was always someone else's job. That gap is what keeps you from building something end-to-end.

Solo Unicorn Builder fills that gap with AI. You bring the expertise in your domain. AI covers the rest — so you can finally ship the whole thing, not just your piece of it.

Don't let the word "coding" scare you. In an AI-native workflow, your daily conversation language — natural language — is the new code.

Your AI-Augmented Team

The 36 core skills are organized as an "Office" — each role is an AI-powered expert you can call on:

Role What it does for you Example skills
CTO Architect, build, test, deploy multi-file-architecture, test-first-development, docker-expert, mcp-builder, webapp-testing
CPO Define what to build and for whom product, idea-validation, pm-design-thinking, frontend-ui-ux
CMO Position, brand, reach your audience marketing-brand, go-to-market, growth-analytics, generative-art
CRO Grow revenue, build partnerships sales, business-development, business-model
CFO Manage costs, plan finances finance-accounting, fundraising, aws-cli-architect
COO Run operations, build your career operations, career-advisor, portfolio-strategy, document-creation

You don't need to hire a team. You need to ask the right questions — and let AI handle the execution. Full list of skills →

Add-on Skills

Beyond the 36 core skills, 7 add-on skills extend the platform:

Skill What it does
career-advisor Full-lifecycle career management — self-discovery, gap analysis, resume generation, interview prep, onboarding, achievement tracking, and self-reviews. Replaces the former career-resume skill.
startup-explorer Pre-validation startup idea explorer — bridges career strengths to market opportunities before you commit to building
notebooklm Query Google NotebookLM notebooks directly from your coding agent
youtube-knowledge-extractor Extract key ideas, summaries, and actionable insights from YouTube videos
oci-expert Oracle Cloud Infrastructure expertise — Always Free tier, A1 ARM64 instances, networking, and common errors
landing-page-service-discovery Synthesize professional accomplishments into high-converting landing page copy
review Lightweight code review skill for bugs, security issues, performance, and readability; adapted from LangChain Deep Agents.

Add-on skills work exactly like core skills — just describe what you need and the AI applies the right one. Want to contribute your own? How to contribute →

Why This Exists

Here's the real problem:

  • Specialization created blind spots. You've spent years going deep in one area. But shipping a product requires breadth — requirements, architecture, testing, deployment, marketing, legal, finance. Nobody taught you the other 80%.
  • AI can fill the gaps — but only with structure. A chatbot can answer questions. It can't guide you through a product launch, a deployment pipeline, or a fundraising round — unless it has a framework for each one.
  • Your thinking is scattered. A thread in ChatGPT, a conversation in Claude, notes in one app, code in another. Nothing connects.

Solo Unicorn Builder gives you 36 core structured skills, 7 add-on skills, a knowledge vault, and a project workspace — so AI can operate as your team across every function you've never done before.

What You Get

When you clone this project and go through this process, you get:

  • Bootstrap your AI agent command center — 43 AI-powered skills: 36 core skills plus 7 add-ons. Not just coding: product development, sales, marketing, legal, finance, operations, code review, and more. Just describe what you need in natural language and the AI applies the right expertise. Full list →
  • A private knowledge vault — Your ideas, notes, goals, technical decisions, and learning in one place. Never checked into the public repo. Ships with template_knowledge/ as a starter — copied to your private my_knowledge/ on init.
  • Starter projects you can build ontemplate_projects/ ships with example projects (like the landing page template) so you're not starting from zero. Your own work lives in my_projects/ — build and ship real projects with AI-assisted workflows. A "project" can be anything: a blog post, a marketing research brief, a deployed web application, or an automated workflow.
  • AI agent vendor-agnostic — Works with any CLI coding agent: Claude Code, Gemini CLI, Kiro CLI, Codex CLI, OpenCode, or any tool that reads markdown. No vendor lock-in. Compare agents →
  • Local sandbox execution — Docker Desktop turns natural-language instructions into safe, containerized execution on your own Mac or Windows machine, then carries the same container path toward production. Why it matters →

Project Structure

The Solo Unicorn Builder project utilizes a distinct folder structure to keep your personal work separate from the core command center, making updates and version control smoother.

  • solo-unicorn/ (This Directory): This is the core command center containing all the AI-powered skills, instructions, templates (template_knowledge/, template_projects/), and project configuration. It's designed to be updated frequently with new features and skills.
  • my_knowledge/ (Sibling to solo-unicorn/): This directory is your private knowledge vault. It's where you'll store all your personal notes, research, ideas, and accumulated context. It's initialized from template_knowledge/ but is intended for your exclusive use and version control.
  • my_projects/ (Sibling to solo-unicorn/): This directory is your workspace for building and shipping projects. You'll copy starter projects from solo-unicorn/template_projects/ into my_projects/ and manage them with their own version control.

This separation ensures that updates to the solo-unicorn/ command center don't interfere with your ongoing projects and personal knowledge base.

Start Here

1. Check Prerequisites

Before cloning, make sure you have the required tools. Read the Prerequisites Guide →

TL;DR: Install Docker Desktop. Everything else runs inside a container.

2. Clone and Set Up

# Clone the project
git clone https://github.com/pingwu/solo-unicorn.git

3. Start Your AI Coding Agent

Start your CLI coding agent from the parent workspace that now contains solo-unicorn/, then tell it: "Run the init unicorn setup" — it will create your personal knowledge vault and connect the skills.

Your first three moves:

  1. "Build me a personal landing page" → ship your first project using AI-paired development
  2. "Help me validate this idea" → the AI uses idea-validation and product to pressure-test before you build
  3. Contribute back → participate in the Solo Unicorn Builder project itself. Contributing to an open-source AI-native platform proves you can ship across multiple roles.

Or follow the manual setup in INIT_UNICORN.md.

Grow With Your Ambition

Solo Unicorn Builder meets you where you are — and scales with you.

Stage You're saying... How it helps...
Learning the gaps "I've never done deployment / product / marketing" AI walks you through each discipline with structured skills
Building end-to-end "I want to ship something from scratch to production" Architecture, testing, debugging, CI/CD, deployment
Launching "I have something — now I need users" Validate ideas, define your product, go to market
Growing "I need to understand the business side" Revenue models, partnerships, legal, finance, operations
Getting hired "I want to prove I can build end-to-end with AI" Portfolio projects, resume tailoring, GitHub presence

From Context Engineering to Agent Harness

The industry calls it context engineering — structuring information so AI produces better outputs. We've been doing it since before the term existed. But context engineering is only half the picture. Knowing what to feed the AI doesn't help if you don't have a system to make it act.

Solo Unicorn Builder is an agent harness — the evolution from context engineering. It doesn't just structure your knowledge; it gives AI agents the skills, instructions, and domain expertise to operate as your team across every function you've never done before.

Three layers make this work:

  • Skills (skills/) — structured prompting patterns that give the AI domain expertise across engineering, product, marketing, sales, finance, legal, and operations
  • Instructions (CLAUDE.mdUNICORN_CONSTITUTION.md) — cascading rules that shape agent behavior, from mission down to individual skill
  • Knowledge (my_knowledge/) — your daily writing, research, ideas, and goals — the layer only you can build

The knowledge layer is the foundation. A daily practice of collecting thoughts, capturing external knowledge, and organizing it into folders that form a personal knowledge graph. Over time, the AI doesn't just answer questions — it reasons about your specific situation, because your accumulated context is always available.

Because your context is plain files — not locked inside any vendor — you can run the same knowledge, the same skills, and the same instructions through different models. Ask Claude Code to architect a feature, then ask Gemini CLI to review it, then ask Kiro CLI to deploy it. Same context, multiple perspectives. That's something you can't do when your history is trapped in a single provider's chat threads.

Proof It Works

This isn't theory. The agent harness approach has shipped real outcomes:

  • Just Ask — A business novel produced entirely through the daily writing and knowledge collection practice that became my_knowledge/. Years of daily reflections, research notes, and accumulated context gave the AI something real to work with. The protagonist discovers the core insight: the technology is never the variable; the context is.
  • AI Launchpad Cohort — A production landing page for a paid 3-week intensive, built and shipped using the same skills and workflows this project teaches. From copy to deployment, every piece was created through the agent harness — proving the framework works for real business outcomes, not just side projects.
  • WRITITATION™ — before the Karpathy Wiki Pattern — In April 2026, Karpathy's LLM Wiki Pattern went viral: structured markdown + Obsidian + LLM agents. Solo Unicorn Builder has been doing this since 2025 — with a trademarked methodology (WRITITATION™) and the obsidian-knowledge skill — over a year before the pattern had a name.

Read the full story →

The Things AI Can't Do for You

  1. Ask your Why — What is your core value? What are your ultimate goals?
  2. Listen to your audience, build trust — Whether 1:1 or 1:many.
  3. Do and implement with meaningful impact — Without 1 and 2, building is just busyness.

AI handles the heavy lifting. You handle the asking, listening, and trust-building. Read more →

Contribute Your Skills

This is a community project. If you've solved a real problem for real people, package it as a skill and share it. How to contribute →

Learn More

Attribution

  • Framework based on practical context engineering and AI-augmented knowledge work patterns.
  • Skills architecture inspired by Anthropic's knowledge-work-plugins — an open-source framework for AI-augmented knowledge work.

License

MIT

About

Unicorn Project - AI-assisted product incubator with container-first development

Resources

Code of conduct

Contributing

Stars

12 stars

Watchers

0 watching

Forks

Releases

Sponsor this project

Packages

Contributors

Languages