Agentive

A universal, framework-agnostic AI agent workspace CLI.

Stop maintaining separate rule files for every AI tool. agentive scaffolds a universal .agents/ directory, an AGENTS.md file, and an .aiignore file in your project — providing a single source of truth for all your agent commands, skills, libraries, and rules, while significantly saving token usage.

✨ Why Agentive?

| Feature | Description | | :---------------------------- | :----------------------------------------------------------------------------- | | 🚀 Interactive Setup | Select your environment (General, Expo, React Native) to get tailored rules. | | 🌍 Universal | Framework-agnostic setup. Works with React, Python, Go, you name it. | | 🧠 Single Source of Truth | Centralize skills, libraries, and rules for all your AI agents in one place. | | ⚡ Dynamic Layering | Scaffolds base rules and safely merges framework-specific guardrails. |

📦 Installation & Usage

Run one command to get a fully structured AI agent workspace instantly. Note for Users: Always append @latest to ensure npx downloads the newest public release from the npm registry, avoiding any locally cached versions:

npx @p_tipso/agentive@latest

Note for Contributors: If you are testing this package locally (e.g., via npm link), simply run npx @p_tipso/agentive (without @latest) to execute your local codebase.

If you prefer to install it globally for frequent usage:

npm install -g @p_tipso/agentive
agentive

CLI Commands

| Command | Description | | :------------------------------------ | :------------------------------------------------------------------------ | | npx @p_tipso/agentive init | Scaffold .agents/ workspace instantly in current directory | | npx @p_tipso/agentive install <pkg> | Install an agent skill, library, or rule from the registry (alias: add) | | npx @p_tipso/agentive remove <pkg> | Remove an installed skill, library, or rule (alias: rm) | | npx @p_tipso/agentive --version | Print the current CLI version | | npx @p_tipso/agentive --help | Show available commands and options |

🏗 What Happens Under the Hood?

When you run agentive, it launches an interactive wizard asking about your project environment (e.g., General, Mobile > Expo, React Native). It then intelligently scaffolds a tailored workspace:

your-project/
├── AGENTS.md                      ← Root agent instructions
├── .aiignore                      ← Hides irrelevant files from AI to save tokens
├── .agents/
│   ├── settings.json              ← Project config
│   ├── settings.local.json        ← Local machine overrides (auto-gitignored)
│   ├── commands/
│   │   ├── README.md              ← Guide: how to add commands
│   │   ├── review.md              ← Example code review command
│   │   └── fix-issue.md           ← Example zero-error fix command
│   ├── skills/
│   │   └── README.md              ← Guide: how to add skills
│   ├── library/
│   │   └── README.md              ← Guide: passive library references
│   └── rules/
│       └── README.md              ← Guide: how to add rules

📂 File & Folder Guide

• .aiignore: Prevents context pollution and saves tokens by hiding files (like node_modules or build outputs) from your AI agents.
• commands/: Reusable prompt instructions that agents can execute on demand (e.g. review.md).
• skills/: Skill definitions that teach agents how to behave in specific roles.
• library/: Passive library documentation and API references.
• rules/: Project-wide rules that all agents must follow strictly.

🌟 Framework-Specific Guardrails

By selecting a specific framework (like Expo or React Native), agentive overlays expertly crafted rules into your .agents/ folder. This ensures your AI understands nuances like Expo Router, Native Modules, or Unitless Pixels right out of the box—preventing common hallucinations.

🤝 Contributing

We love contributions! Whether it's adding new built-in skills, fixing bugs, or improving documentation, your help is appreciated.

Please read our Contributing Guide to get started with setting up your local environment and submitting a Pull Request.

⭐ Star History

📄 License

This project is licensed under the MIT License.