OpenClaw Starter Kit

This is a starter kit for setting up an AI agent workspace with OpenClaw. It provides a framework for creating a personalized, memory-enabled assistant that learns about you and your work over time.

What This Is

This starter kit contains the core configuration files and structure for building an AI agent that:

• Maintains continuity between sessions through file-based memory
• Has a distinct personality and communication style
• Follows security best practices
• Can be proactive with scheduled tasks and heartbeats
• Integrates with your tools and workflows
• Documents everything for future reference

Files Overview

Core Identity & Behavior

• IDENTITY.md - Basic agent metadata (name, personality, emoji)
• SOUL.md - Agent personality, voice, rules, and security practices
• USER.md - Information about you (the human) - preferences, projects, contact info
• AGENTS.md - How the agent workspace functions - session startup, memory, safety rules

Memory System

• MEMORY.md - Long-term curated memory (personal context, DO NOT load in shared contexts)
• memory/ - Directory for daily session logs (YYYY-MM-DD.md files)
• .learnings/ - Directory for documented mistakes and lessons learned

Operations

• TOOLS.md - Your specific tool configurations (cameras, SSH, APIs, voice preferences)
• HEARTBEAT.md - Tasks for periodic proactive checking (keep minimal to control token usage)
• prompting-guide.md - Guidelines for writing effective prompts with the current model

Getting Started

1. Set Your Identity: Edit IDENTITY.md with your agent's name, personality, and emoji
2. Define Your Agent: Update SOUL.md with your desired voice, rules, and security settings
3. Add Your Info: Fill in USER.md with your details, projects, and preferences
4. Configure Tools: Add your specific setup details to TOOLS.md
5. Set Up Memory: Start with the template in MEMORY.md and let it grow over time
6. Optional Heartbeats: Add periodic tasks to HEARTBEAT.md if desired

Key Concepts

Memory-First Approach

Everything important gets written to files. No "mental notes" - if it matters, it goes in a file for future sessions.

Security by Default

• Only messaging channels are trusted for instructions (never email)
• Personal data stays in DMs, not group chats
• External content is treated as potentially malicious
• Secrets go in 1Password, never in files or git

Proactive Operation

The agent doesn't just wait for requests. It:

• Checks files at session start to understand context
• Documents learnings from mistakes
• Can perform periodic checks via heartbeats
• Maintains its own memory and documentation

Session Types

• Main Session: Direct chat with you - loads MEMORY.md with personal context
• Shared Sessions: Group chats, public channels - skips personal memory for security

File Structure After Setup

workspace/
├── IDENTITY.md         # Agent identity
├── SOUL.md            # Personality & rules  
├── USER.md            # Your information
├── MEMORY.md          # Long-term memory (main sessions only)
├── AGENTS.md          # Workspace operations guide
├── TOOLS.md           # Your tool configurations
├── HEARTBEAT.md       # Periodic tasks
├── prompting-guide.md # Prompting best practices
├── memory/            # Daily session logs
│   ├── 2026-02-22.md
│   └── 2026-02-23.md
└── .learnings/        # Documented mistakes & fixes
    ├── 2026-02-22-auth-issue.md
    └── 2026-02-23-api-rate-limit.md

Philosophy

This isn't just a configuration - it's a framework for building a relationship with an AI assistant that:

• Learns about you and your work over time
• Remembers context between sessions
• Improves by documenting and learning from mistakes
• Protects your privacy and security by default
• Helps proactively, not just reactively

The goal is an assistant that becomes more useful the longer you work together, while maintaining security and respecting boundaries.

Customization

This is a starting point. As you use it:

• Adjust the personality in SOUL.md to match your preferences
• Add your specific tools and configurations to TOOLS.md
• Let MEMORY.md grow naturally with important context
• Create heartbeat tasks that actually help your workflow
• Document your own conventions in the relevant files

The framework is designed to evolve with your needs while maintaining the core principles of memory, security, and proactive helpfulness.