From rcc
Bootstraps new projects from scratch: gathers requirements for frontend SPAs, full-stack apps, APIs, CLI tools, libraries; researches latest best practices via WebSearch; creates blueprint; confirms; initializes with official CLI; sets up Claude Code agent system.
npx claudepluginhub wayne930242/reflexive-claude-codeThis skill uses the workspace's default tool permissions.
**Initializing projects IS bootstrapping with modern best practices AND agent system.**
Initializes Claude Code for new projects: runs skeleton script for directories/boilerplate, interactively generates AGENTS.md sections via user prompts, optional research profile. Triggered by 'init project'.
Generates complete production-ready projects from descriptions: directory structure, code, tests, docs, config, git init. Orchestrates pipeline with verification to ensure builds and tests pass.
Guides beginners through pain point identification to suggest and set up their first agentic Claude Code project. Use for 'what project should I build?' or getting started queries.
Share bugs, ideas, or general feedback.
Initializing projects IS bootstrapping with modern best practices AND agent system.
Don't just create files—set up the full development environment including Claude Code integration.
Core principle: Research current best practices. Don't assume—verify with official docs.
Violating the letter of the rules is violating the spirit of the rules.
Pattern: Chain
Handoff: user-confirmation
Next: brainstorming-workflows
Chain: bootstrap
Before ANY action, create task list using TaskCreate:
TaskCreate for EACH task below:
- Subject: "[initializing-projects] Task N: <action>"
- ActiveForm: "<doing action>"
Tasks:
Announce: "Created 7 tasks. Starting execution..."
Execution rules:
TaskUpdate status="in_progress" BEFORE starting each taskTaskUpdate status="completed" ONLY after verification passesTaskList to confirm all completedGoal: Understand what project to create.
First, ask Project Type:
| Project Type | Follow-up Questions |
|---|---|
| Frontend SPA | Language, Framework, UI Library, Styling, State, Testing |
| Full-stack App | Above + ORM, Database, API Style, Auth |
| API Service | Language, Framework, ORM, Validation, Auth |
| CLI Tool | Language, Package Manager, Testing |
| Library | Language, Build Tool, Testing, Docs |
Ask iteratively. Skip already-answered or N/A questions.
Verification: Have answers to all relevant questions for project type.
Goal: Find current best practices for the chosen stack.
Use WebSearch to find:
CRITICAL: Don't assume. Official docs change frequently.
Verification: Have official init command and current version numbers.
Goal: Create a blueprint document for user review.
Write to: /docs/blueprint.md
# Project Blueprint
## Stack
- Language: [name] [version]
- Framework: [name] [version]
- Package manager: [choice]
## Initialization Command
\`\`\`bash
[official CLI command]
\`\`\`
## Project Structure
[Expected directory layout after init]
## Key Dependencies
[Core packages to add]
## Agent System Plan
- CLAUDE.md: [key laws]
- Rules: [planned rules]
- Hooks: [planned quality gates]
Verification: Blueprint is written and covers all sections.
Goal: Have user approve the blueprint before execution.
CRITICAL: Do not proceed without explicit user confirmation.
Present the FULL blueprint content to user. Do NOT summarize — show every section with detail:
Anti-pattern: Listing 3-5 bullet points of one-liners is NOT presenting. The user must see enough detail to catch wrong versions, missing dependencies, or incorrect structure.
Ask: "這個 blueprint 正確嗎?要開始執行嗎?"
Verification: User has reviewed the full blueprint and explicitly approved.
Goal: Run the official CLI to create the project.
Process:
Example:
npx create-next-app@latest my-app --typescript --tailwind --eslint
cd my-app && npm run dev
Verification: Project builds and runs without errors.
Goal: Configure Claude Code for the new project.
Since this is a new project, skip analysis and go directly to workflow exploration.
Handoff: "專案已建立並通過驗證。要繼續設定 agent system 嗎?"
brainstorming-workflows skillThe skill chain will automatically continue:
brainstorming-workflows → explore user's workflowsplanning-agent-systems → plan componentsapplying-agent-systems → create components via writing-* skillsreviewing-agent-systems → quality review with all reviewer agentsrefactoring-agent-systems → fix issues from reviewVerification: Agent system is configured and verified by the skill chain.
Goal: Verify everything works together.
Checklist:
Verification: All checklist items pass.
npx create-next-app@latest --typescript --tailwind
# or
npm create vite@latest -- --template react-ts
npx create-next-app@latest --typescript
# Add: prisma, next-auth, zod
npm init -y && npm install express typescript
# or
cargo new --name api
npm init -y && npm install commander
# or
cargo new --name cli
These thoughts mean you're rationalizing. STOP and reconsider:
All of these mean: You're about to create a weak foundation. Follow the process.
| Excuse | Reality |
|---|---|
| "I know the version" | Versions change monthly. Verify. |
| "Skip research" | Best practices evolve. Check official docs. |
| "Skip confirmation" | Blueprint approval prevents wasted effort. |
| "Brief summary is enough" | Brief summaries hide wrong decisions. Show full detail. |
| "Agent system later" | Set it up now while context is fresh. |
| "Fresh = working" | Fresh projects can have config issues. Validate. |
digraph init_project {
rankdir=TB;
start [label="New project", shape=doublecircle];
gather [label="Task 1: Gather\nrequirements", shape=box];
research [label="Task 2: Research\nbest practices", shape=box];
blueprint [label="Task 3: Write\nblueprint", shape=box];
confirm [label="Task 4: User\nconfirmation", shape=box];
confirmed [label="Approved?", shape=diamond];
bootstrap [label="Task 5: Bootstrap\nproject", shape=box];
agent [label="Task 6: Setup\nagent system", shape=box];
validate [label="Task 7: Final\nvalidation", shape=box];
valid [label="All\npasses?", shape=diamond];
done [label="Project ready", shape=doublecircle];
start -> gather;
gather -> research;
research -> blueprint;
blueprint -> confirm;
confirm -> confirmed;
confirmed -> bootstrap [label="yes"];
confirmed -> blueprint [label="no\nrevise"];
bootstrap -> agent;
agent -> validate;
validate -> valid;
valid -> done [label="yes"];
valid -> bootstrap [label="no\nfix"];
}
| Step | Skill | Purpose |
|---|---|---|
| 1 | brainstorming-workflows | Role-based workflow exploration |
| 2 | planning-agent-systems | Component planning |
| 3 | applying-agent-systems | Invoke writing-* skills |
| 4 | reviewing-agent-systems | Quality review with reviewer agents |
| 5 | refactoring-agent-systems | Fix issues from review |