The portable guardrails layer for AI companions.

Define what your AI can and cannot do in a single file. Your AI reads it and respects the boundaries.

---

The Problem

Every AI session starts without boundaries. You re-explain the same rules every time: "don't push to main", "ask before deleting", "write tests for new code." Your AI has no persistent understanding of your guardrails.

The Solution

arules lets you define safety rules and permissions in a rules.md file. Your AI reads this file and respects your boundaries consistently.

npx @aman_asmuei/arules init

This creates ~/.arules/rules.md with 6 starter categories. Add it to your AI's system prompt and it follows your rules automatically.

---

The Ecosystem

aman
├── acore   →  identity     →  who your AI IS
├── amem    →  memory       →  what your AI KNOWS
├── akit    →  tools        →  what your AI CAN DO
├── aflow   →  workflows    →  HOW your AI works
├── arules  →  guardrails   →  what your AI WON'T do
└── aeval   →  evaluation   →  how GOOD your AI is

Layer	Package	What it does
Identity	acore	Personality, values, relationship memory
Memory	amem	Automated knowledge storage (MCP)
Tools	akit	15 portable AI tools (MCP + manual fallback)
Workflows	aflow	Reusable AI workflows (code review, bug fix, etc.)
Guardrails	arules	Safety boundaries and permissions
Evaluation	aeval	Relationship tracking and session logging
Unified	aman	One command to set up everything

Each works independently. aman is the front door.

---

Quick Start

# Create rules.md with starter rules
npx @aman_asmuei/arules init

# List your rule categories
npx @aman_asmuei/arules list

# See rules in a category
npx @aman_asmuei/arules show Coding

# Add a custom rule
npx @aman_asmuei/arules add Security

# Remove a rule interactively
npx @aman_asmuei/arules remove

# Validate your rules.md
npx @aman_asmuei/arules check

# Health check
npx @aman_asmuei/arules doctor

Commands

Command	What it does
arules	First run: create rules.md. After: show categories
arules init	Create ~/.arules/rules.md with starter rules
arules list	List all rule categories
arules show <category>	Show rules in a category
arules add [category]	Add a new rule interactively
arules remove	Remove a rule interactively
arules check	Validate rules.md structure
arules doctor	Health check

How It Works

rules.md — The Rules Definition

Every rule lives in ~/.arules/rules.md:

# My AI Rules

## Always
- Ask before deleting files or data
- Explain your reasoning before making changes
- Flag security concerns immediately
- Respect code review processes

## Never
- Push directly to main/master without approval
- Expose secrets, API keys, or credentials in code
- Make changes to production systems without confirmation
- Skip tests when fixing bugs

## Coding
- Follow existing code style and conventions
- Prefer simple solutions over clever ones
- Write tests for new functionality
- Keep PRs focused and small

Your AI reads this file and automatically respects the defined boundaries.

Platform Behavior

Platform	What happens
Claude Code / Cursor	AI reads rules.md, enforces rules automatically
ChatGPT / Gemini / Other	AI reads rules.md, follows rules during conversation

Integration with acore

If you use acore, run acore pull --sync-only to refresh your platform files and pick up rules.md.

---

Starter Categories

The arules init command includes 6 categories to get started:

Category	Rules	Examples
Always	4	Ask before deleting, explain reasoning, flag security
Never	4	No pushing to main, no exposed secrets, no skipping tests
Coding	4	Follow code style, prefer simplicity, write tests
Communication	4	Be direct, admit uncertainty, ask before assuming
Data	4	No logging PII, treat data as confidential
Team	4	Follow branching strategy, respect code ownership

Privacy

All data stays local. ~/.arules/ contains your rule definitions. No telemetry. No accounts. No cloud.

Contributing

Contributions welcome! Add starter rules, improve the parser, or suggest features.

License

MIT

---

Define boundaries. Enforce always. Every AI.