Skip to content
/ .ai Public

Framework para desarrollo de software con agentes IA. Contexto persistente, estándares de código y flujo metódico que evita pérdida de contexto entre sesiones.

License

Notifications You must be signed in to change notification settings

soydachi/.ai

Repository files navigation

AI Way of Working (ai-wow)

Single Source of Truth for AI-assisted software engineering.

PyPI version CI Python 3.9+ License: MIT

This framework provides a standardized, tool-agnostic structure for working with AI coding assistants. It integrates seamlessly with GitHub Copilot, Claude, and other AI tools while maintaining consistency across projects.

🚀 Installation

Option 1: pip (Recommended)

pip install ai-wow

Option 2: One-liner Installation

Linux/macOS:

curl -sSL https://raw.githubusercontent.com/soydachi/ai-wow/main/install.sh | bash

Windows (PowerShell):

iwr -useb https://raw.githubusercontent.com/soydachi/ai-wow/main/install.ps1 | iex

Option 3: From Source

git clone https://github.com/soydachi/ai-wow.git
cd ai-wow
pip install -e .

Option 4: GitHub Template

Use this repository as a template to include the full .ai/ framework structure:

  1. Click "Use this template" on GitHub
  2. Create your new repository
  3. The .ai/ structure is ready to customize

Option 5: Git Submodule (Advanced)

For teams wanting independent framework updates:

# Add as submodule
git submodule add https://github.com/soydachi/ai-wow.git .ai-framework
ln -s .ai-framework/.ai .ai

# Update to latest
git submodule update --remote

📖 Quick Start

# Initialize in a new project
ai-wow init --name "MyProject" --stack dotnet

# Sync to GitHub Copilot
ai-wow sync github

# Sync to Claude
ai-wow sync claude

# Validate framework structure
ai-wow validate

# Update to latest version
ai-wow update

🛠️ CLI Commands

ai-wow init

Initialize the .ai/ framework in a new project.

ai-wow init --name "MyAPI" --stack dotnet
ai-wow init --name "WebApp" --stack typescript --target ./my-project
ai-wow init -n "DataPipeline" -s python --include-examples

Options:

  • --name, -n - Project name (required)
  • --stack, -s - Technology stack: dotnet, typescript, python, terraform, multi (required)
  • --target, -t - Target directory (default: current)
  • --include-examples, -e - Include example files
  • --force, -f - Overwrite existing .ai/ directory

ai-wow sync

Synchronize .ai/ content to AI tool configurations.

ai-wow sync github          # Sync to GitHub Copilot (.github/)
ai-wow sync claude          # Sync to Claude (.claude/)
ai-wow sync all             # Sync to all tools
ai-wow sync github --dry-run  # Preview changes

Options:

  • --force, -f - Overwrite existing files
  • --dry-run, -n - Preview without making changes

ai-wow validate

Validate the .ai/ framework structure and content.

ai-wow validate           # Basic validation
ai-wow validate --fix     # Validate and fix issues
ai-wow validate -v        # Verbose output

Options:

  • --fix, -f - Attempt to fix issues
  • --verbose, -v - Show detailed results

ai-wow update

Update ai-wow to the latest version.

ai-wow update              # Update CLI to latest
ai-wow update --check      # Check for updates only
ai-wow update --framework  # Update CLI and project templates

Legacy Script Support

For backwards compatibility, you can still use Python scripts directly:

python .ai/tools/sync_github.py
python .ai/tools/sync_claude.py
python .ai/tools/validate.py

📁 Framework Structure

.ai/
├── README.md                    # This file
├── config.yaml                  # Global configuration
│
├── context/                     # 🎯 PROJECT CONTEXT
│   ├── project.md               # Vision, objectives, stakeholders
│   ├── architecture.md          # System architecture (C4 compatible)
│   ├── stack.md                 # Technology stack details
│   ├── glossary.md              # Domain terminology
│   └── decisions/               # Architecture Decision Records (ADRs)
│
├── standards/                   # 📏 CODING STANDARDS
│   ├── _index.yaml              # Standards registry
│   ├── global.md                # Cross-stack rules
│   ├── dotnet/                  # .NET specific
│   ├── typescript/              # TypeScript/React specific
│   ├── python/                  # Python specific
│   └── infrastructure/          # Terraform/IaC specific
│
├── prompts/                     # 💬 REUSABLE PROMPTS
│   ├── _index.yaml              # Prompts registry
│   ├── system.md                # Base system prompt
│   └── templates/               # Prompt templates by use case
│
├── skills/                      # 🔧 MODULAR SKILLS
│   ├── _index.yaml              # Skills registry
│   ├── dotnet/                  # .NET skills
│   ├── typescript/              # TypeScript skills
│   └── cross-cutting/           # Universal skills
│
├── agents/                      # 🤖 AUTONOMOUS AGENTS
│   ├── _index.yaml              # Agents registry
│   ├── feature-builder/         # Multi-step feature creation
│   ├── code-reviewer/           # Code review automation
│   └── migrator/                # Version migration assistance
│
├── learnings/                   # 📚 EVOLUTIONARY LEARNINGS
│   ├── global.md                # Cross-project learnings
│   └── by-stack/                # Stack-specific learnings
│
└── tools/                       # 🛠️ AUTOMATION SCRIPTS (Legacy)
    ├── sync_github.py           # Wrapper for ai-wow sync github
    ├── sync_claude.py           # Wrapper for ai-wow sync claude
    └── validate.py              # Wrapper for ai-wow validate

🔌 AI Tool Integration

GitHub Copilot

ai-wow sync github generates:

  • .github/copilot-instructions.md - Main instructions from prompts/system.md
  • .github/instructions/*.md - Stack-specific instructions from standards/
  • .github/prompts/*.md - Prompt templates

Claude

ai-wow sync claude generates:

  • .claude/CLAUDE.md - Combined project documentation
  • .claude/commands/*.md - Prompt templates as commands
  • .claude/docs/*.md - Stack-specific standards

Other Tools

The agnostic .ai/ structure can be adapted to any AI tool by creating a new sync command.

🧩 Core Concepts

Hierarchy: Prompts → Skills → Agents

Concept Hierarchy

Concept Purpose Autonomy Example
Prompt Define AI behavior None (static) "Use Result pattern"
Skill Atomic capability Low (invoked) "Create API endpoint"
Agent Orchestrate workflows High (autonomous) "Build complete feature"

When to Use What

Task Use
Change base behavior Prompt
Repetitive atomic task Skill
Complex multi-step work Agent

🌍 Cross-Platform Support

ai-wow works on:

  • Windows 10/11 (PowerShell 5.1+, CMD)
  • macOS (Intel & Apple Silicon)
  • Linux (Ubuntu, Debian, Fedora, Arch, etc.)

Requirements:

  • Python 3.9 or higher
  • pip (Python package manager)

📝 File Format Conventions

YAML Front Matter (for tool integration)

---
id: skill-id
name: Human Readable Name
description: Brief description for AI tools
applyTo: "**/*.cs"              # Glob pattern (optional)
tags: [dotnet, api]
---

Learnings Format

* Learning description with context (weight)
  • Weight starts at 1, increases with confirmation
  • Higher weight = higher priority for AI consideration

ADR Format (Architecture Decision Records)

# ADR-001: Title

## Status
Accepted | Superseded | Deprecated

## Context
Why this decision was needed

## Decision
What was decided

## Consequences
Impact of the decision

🤝 Contributing

  1. Fork the repository
  2. Create your feature branch (git checkout -b feature/amazing-feature)
  3. Follow existing patterns when adding new content
  4. Update _index.yaml files when adding skills/agents
  5. Run ai-wow validate before committing
  6. Submit a Pull Request

📄 License

MIT - See LICENSE file for details.

🔗 Links

About

Framework para desarrollo de software con agentes IA. Contexto persistente, estándares de código y flujo metódico que evita pérdida de contexto entre sesiones.

Topics

Resources

License

Contributing

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Contributors 2

  •  
  •