ConstitutionKit

Preamble

The founding document
for your AI-built codebase.

Draft a production-ready constitution.md, CLAUDE.md, or AGENTS.md — the record every engineer and every agent on the project is bound to. Tailored to your stack, your team, your principles. Runs entirely in your browser: no account, no server.

spec-kit Claude Code AGENTS.md agentic AI no account needed open source · Apache-2.0

Article I

Choose your instrument

spec-kit

→ constitution.md

Foundational principles and behavioral contracts for the whole project. Read by humans and AI agents alike.

Claude Code

→ CLAUDE.md

Agent-specific instructions: allowed tools, forbidden paths, project commands, and autonomy boundaries.

Both

→ constitution.md + CLAUDE.md

Generate both files. CLAUDE.md will reference the constitution rather than duplicate it.

AGENTS.md

→ AGENTS.md

The cross-tool standard read by Codex, Cursor, Copilot, and other coding agents. Pick this alongside Claude Code to get a thin, @-importing CLAUDE.md instead of a duplicate file.

Article II

Classify the project

🌐

Web App

SPA / SSR / fullstack web product

🔗

Microservices

Distributed services with APIs

🔄

Data Pipeline

ETL, streaming, analytics, ML

🤖

AI System

Agentic, LLM-powered, or AI-first

⚙️

Platform / SDK

Developer tooling, libraries, APIs

📱

Mobile App

iOS, Android, or cross-platform

🐦

Flutter App

Cross-platform Flutter / Dart app

🏗️

Full-Stack Solution

Multi-tier: mobile, web, API, infra

💬

WhatsApp Agent

AI-powered WhatsApp bot via Business API

Select tiers in your solution (pick all that apply)

📱 Mobile App

🌐 Web / Admin App

⚙️ Backend API

🔗 Microservices

🗄️ Database / Cache

🔐 Auth Service

Tech stack (pick all that apply)

Frontend

Next.js

React

Vue

TypeScript

Backend & languages

Node.js

Python

Rust

Java / Spring

Data & APIs

PostgreSQL

MongoDB

Redis

Kafka

GraphQL

tRPC

Infra & cloud

Docker

Kubernetes

AWS

GCP

Terraform

Mobile

React Native

Flutter

Dart

Swift

Kotlin

iOS

Android

Firebase

AI / LLM

LangChain

Anthropic API

Messaging & webhooks

WhatsApp API

Twilio

Meta Cloud API

Webhook

Article III

Enumerate the principles

Quick presets

Web App Essentials

AI-First Stack

Startup Speed

Enterprise Grade

Security-Critical

Microservices

Data Pipeline

Platform / SDK

Mobile (Native)

Flutter / Dart

Full-Stack

WhatsApp Agent

Everything

Clear all

0 selected

Article IV

Record the particulars

Project description — optional

Commands & structure

Build / test commands

Used in CLAUDE.md so the agent knows how to run the project

Forbidden paths

Directories the agent must never touch autonomously

Key directories

Important folders in your project structure

Test framework & coverage threshold

e.g. Jest + RTL, 80% coverage / pytest, no threshold

Team & workflow

Agent autonomy level

How much the agent can do without asking

Team size

Influences collaboration and review conventions

Branch strategy

How your team manages branches and merges

PR merge strategy

How PRs are merged into the target branch

Safety & environment

Environment topology

List your environments and what the agent is allowed to do in each

Unusual constraints or non-standard decisions

What would surprise a new engineer or AI agent? Legacy modules, non-standard patterns, historical context.

100% browser-side. No API, no account. Edit the output before downloading.

The record

Edit directly in the text field, then download.