shinpr/ai-coding-project-boilerplate
shinpr/ai-coding-project-boilerplate has 239 GitHub stars and 28 forks. GitHot measured 0 stars in 24 hours, +3 in 7 days, and +3 in 15 days. Last checked 2026-11-16 23:44 UTC. Last push 2026-11-13.
What you can do
- Develop with TypeScript, Biome formatting and linting, and Vitest already configured
- Record project context and quality standards for later sessions and other contributors
- Turn team knowledge into skills Claude can use when the work calls for them
Measured star growth
- in 24h
- 0
- in 7 days
- +3
- in 15 days
- +3
- in 1 month
- Update delayed
Net star changes over overlapping periods, not daily totals. Missing measurements are left blank. Stars measure attention, not software quality.
About
Agentic coding TypeScript boilerplate for Claude Code: sub-agent workflows with built-in quality checks and context engineering.
ProductivityREADME.md
OriginalAI Coding Project Boilerplate: A Starter Kit for Claude Code
Read this in other languages: 日本語 | 简体中文
Set up a TypeScript repository for development with Claude Code. create-ai-project adds a project-level CLAUDE.md, ready-to-use commands, specialized agents, and skills so Claude can follow your repository's rules while working on changes.
Use this starter kit to create a new project and keep its Claude Code setup up to date. You get a working development environment without assembling prompts and agent definitions yourself. Your team can version, share, and adapt that environment alongside the code.
What you can start with
- Develop with TypeScript, Biome formatting and linting, and Vitest already configured
- Record project context and quality standards for later sessions and other contributors
- Turn team knowledge into skills Claude can use when the work calls for them
Quick start
Requires Node.js 26.12.1+, pnpm, and Claude Code.
Start a new project
npx create-ai-project my-project
cd my-project
pnpm install
claude
Add --lang=ja or --lang=zh-CN to the first command to use Japanese or Simplified Chinese workflow instructions.
Run your first change
Once Claude Code is running:
/project-inject
/implement Add rate limiting to the API
/project-inject sets up the initial project context.
/implement clarifies the request and inspects the existing code. Straightforward changes can proceed without design documents. When design decisions are needed, it prepares a design and plan for your approval before implementation. It then runs the applicable checks and reviews larger changes.
See the Quick Start Guide for the full setup and first-run walkthrough.
Choose a workflow
| What you want to do | Start with |
|---|---|
| Take a feature from a request through implementation and verification | /implement |
| Make a focused change with a clear scope | /task |
| Design a change before coding | /design, /front-design |
| Turn an approved design into an executable plan | /plan, /front-plan |
| Implement an approved plan | /build, /front-build |
| Review completed implementation against the agreed outcome and repository standards | /review, /front-review |
| Investigate a problem before choosing a fix, without changing code | /diagnose |
Design now, implement later
/design Add rate limiting to the API
/plan
/build
The design command stops for approval. You can continue planning and implementation later, in a new session or with another contributor, using the approved documents. The plan records what each task must deliver and how to check it, so the handoff does not depend on reconstructing the earlier conversation.
For frontend work, use the corresponding commands:
/front-design Add a user profile dashboard
/front-plan
/front-build
See Use Cases & Commands for examples and the complete command reference.
Workflow modes
Normal Mode is the default. Choose Lite Mode when you want fewer independent checks and agent calls. It skips checks of design documents against the code and against each other, along with the separate security review. For planned work, Normal Mode runs repository checks such as lint and tests after each task. Lite Mode runs them together after implementation, before code review.
Claude still checks its changes, reviews required tests, and asks for your approval at the same points. Code review remains in place. Small changes and fixes made after review still get the usual quality checks.
To save Lite Mode as the project's default, run this from the project root:
node scripts/set-workflow-mode.js lite
The setting is saved in CLAUDE.md, so future sessions use it without a mode request. Run node scripts/set-workflow-mode.js normal to restore Normal Mode. A mode you explicitly request in conversation overrides the project default for the rest of that session, unless you change it again.
Adapt the environment to your project
Use /project-inject to record the project's purpose, constraints, quality standards, conventions, and external sources. Run it again when those facts change.
For guidance that applies only to particular tasks, use /create-skill or /refine-skill. These commands help you decide where the guidance belongs and review it before use. See the Skills Editing Guide for examples.
Keep the Claude Code setup up to date
From the root of a project created with this starter kit, preview the update, then apply it:
npx create-ai-project update --dry-run
npx create-ai-project update
claude
The updater replaces the managed Claude Code rules, commands, agents, and skills. Local edits to files being updated are overwritten. It preserves your source code, existing package settings, and saved workflow mode.
Before updating, commit your project-specific rules and skills, including changes made with /project-inject, /create-skill, or /refine-skill. After the update, review what changed and manually reapply the customizations you want to keep. See the Quick Start Guide for update details.