AI Product Teardown Studio

Break down AI repositories into product structure, AI surfaces, reusable patterns, and originality-safe new skill ideas.

ai-product-teardown-studio is a local Codex skill plus lightweight Python tooling for classifying repositories, mapping how AI actually participates in them, and turning what you learn into a cleaner, more original next project.

Highlights

• Classify before you imitate: distinguish standard skills, AI apps, workflow repos, embedded-skill repos, and portfolio/showcase projects.
• Teardown with structure: map user entry points, modules, state flow, data/storage, deployment, and AI touchpoints.
• AI surface audit: detect CLAUDE.md, AGENTS.md, .claude/skills/, modes/, MCP configs, and runtime AI layers.
• Pattern extraction, not copying: turn repo observations into reusable pattern cards and originality-safe skill concepts.
• Local and analysis-only: scripts inspect local repository content and do not upload it anywhere by themselves.

What it helps you answer

• Is this repository mainly a product, a skill, a workflow system, or a showcase site?
• How is AI actually wired into the user journey and runtime behavior?
• Which parts are reusable patterns, and which parts are repo-specific branding or prompt wording?
• What can I safely learn from this repo without cloning its expression?
• How can several sample repos be reorganized into a stronger original skill?

Why this is different

• It does not confuse application repos with skill repos.
• It looks at both product structure and AI surfaces.
• It includes originality guardrails, not just teardown notes.
• It is designed to help you build a new skill or workflow, not a disguised copy of an old one.

Workflow

flowchart LR
    A["Input repo<br/>GitHub URL or local path"] --> B["Classify repo type<br/>skill / app / workflow / portfolio"]
    B --> C["Map product structure<br/>entry points, modules, state, data, deploy"]
    C --> D["Audit AI surface<br/>docs, skills, modes, MCP, runtime AI"]
    D --> E["Extract reusable patterns<br/>pattern cards, migration hints, guardrails"]
    E --> F["Draft originality-safe new skill<br/>new positioning, new outputs, new boundaries"]

Loading

Repository contents

Path	Purpose
SKILL.md	Main Codex skill instructions and teardown workflow
scripts/repo_inventory.py	Classify repository structure, languages, frameworks, deployment/storage signals
scripts/feature_signal_scan.py	Detect AI/product/workflow signals such as skills, modes, MCP, approval loops, and scraping
scripts/teardown_report_builder.py	Generate a four-part teardown report
references/	Classification rules, teardown method, AI surface audit, reusable patterns, example prompts, sample outputs, FAQ/troubleshooting, release process, originality rules, and output templates
agents/openai.yaml	UI metadata for the skill

Quick start: inspect a repository

Requirements:

• Python 3.9+
• Local access to the repository you want to inspect

git clone https://github.com/sushuqiong/ai-product-teardown-studio.git
cd ai-product-teardown-studio

python .\scripts\repo_inventory.py `
  "C:\path\to\repo\job-tracker"

Scan richer product/AI signals:

python .\scripts\feature_signal_scan.py `
  "C:\path\to\repo\matchbook"

Generate a structured teardown report:

python .\scripts\teardown_report_builder.py `
  "C:\path\to\repo\career-ops"

Example use cases

Situation	Recommended command or prompt	What you should get back
You found a polished AI app repo and want to understand the product	python .\scripts\repo_inventory.py "C:\path\to\job-tracker"	A classification and structural inventory
You want to know where AI is actually wired into the workflow	python .\scripts\feature_signal_scan.py "C:\path\to\matchbook"	AI surface signals, approval loops, runtime clues
You want a clean four-part teardown	python .\scripts\teardown_report_builder.py "C:\path\to\career-ops"	Product summary, AI surface map, reusable patterns, original skill direction
You want Codex to compare two repos	Use $ai-product-teardown-studio Compare these two repositories...	Shared patterns, product philosophy differences, originality-safe reuse ideas

Use as a Codex skill

Clone or copy this repository into your local Codex skills directory, then invoke:

Use $ai-product-teardown-studio Break down this repository, tell me whether it is mainly a product, a skill, or a workflow system, and extract reusable patterns without copying its wording.

Chinese examples:

Use $ai-product-teardown-studio 拆解这个 GitHub 仓库的产品结构、AI 介入点和可复用模式，并告诉我它更像 skill 还是产品。

Use $ai-product-teardown-studio 比较 job-tracker 和 career-ops 的不同产品哲学，重点看 router、profile layer、运行时 AI 和状态流。

Use $ai-product-teardown-studio 从 matchbook 提炼一个 human-in-the-loop approval pattern，并说明它还能迁移到哪些业务场景。

Use $ai-product-teardown-studio 基于 job-tracker、career-ops、alibi-day-tracker 这些样本，给我一个原创性更强的个人 skill 草案。

More examples:

• Example prompts
• Sample teardown output
• FAQ / troubleshooting
• Release process

Sample repository interpretations

• job-tracker → typically an AI application repo
• career-ops → typically a standard skill + agent workflow repo
• alibi-day-tracker → typically an AI application repo with embedded skills
• matchbook → typically a business automation agent project with approval logic

These are sample inputs, not hidden templates or upstream dependencies.

What a strong teardown should contain

A useful teardown is not just a label like “this is an AI app.” It should usually answer:

1. What kind of repo this is
2. How the user enters the product
3. Where state, data, and deployment live
4. Where AI actually changes the workflow
5. Which patterns are reusable
6. Which parts are too repo-specific to copy
7. What a stronger original follow-up skill or product could look like

If the output does not reach those seven points, it is probably still too shallow.

Common failure modes

• Treating every repo with prompts as a skill repo
• Listing frameworks without explaining why they matter to the product
• Confusing documentation AI with runtime AI
• Copying naming, modes, or README positioning too closely
• Skipping originality boundaries when turning insights into a new skill draft

The references folder is meant to help avoid these mistakes.

Privacy and safety

• This repository is designed for local analysis.
• The included scripts inspect repository files and structural signals, but do not upload repository content anywhere by themselves.
• Avoid feeding private repositories, confidential prompts, internal credentials, or protected business logic into any external workflow unless you explicitly intend that.
• This project helps with analysis and skill ideation; it does not provide legal, security, or compliance certification.

Notes on originality

• This project is built to extract patterns, not clone expression.
• It should not be used to rewrite someone else’s README, SKILL.md, or prompts with superficial edits.
• The intended output is a stronger, cleaner, more original next project.

Contributing

Contributions are welcome, especially for better repo classification, stronger AI surface detection, safer originality guardrails, and clearer teardown outputs.

See CONTRIBUTING.md for contribution scope, reporting guidance, and the lightweight release rhythm used by this repository.

Reading path

If you are new to the project, this is the fastest path:

1. Read SKILL.md
2. Read references/intake-classification.md
3. Read references/product-teardown-method.md
4. Skim references/example-prompts.md
5. Skim references/sample-teardown-output.md
6. Read references/faq-troubleshooting.md
7. Skim references/release-process.md
8. Run the scripts on one real repository

Keywords

AI repository analysis · agent workflow · GitHub project teardown · Codex skill · product structure · prompt engineering · skill design