OpenClaw Agent Config
Configuration, personality, and workspace files for an OpenClaw AI agent running on a VPS.
What's Here
| File | Purpose |
|---|---|
SOUL.md | Agent personality and behavioral guidelines |
AGENTS.md | Workspace conventions, memory practices, heartbeat rules |
IDENTITY.md | Agent identity (name, emoji, vibe) |
USER.md | Info about the human (Shawn) |
TOOLS.md | Local tool notes (SSH aliases, cameras, etc.) |
HEARTBEAT.md | Periodic background tasks |
BOOTSTRAP.md | Original setup script |
CONTEXT.md | Start here — notes for a new AI inheriting this setup |
openclaw.json | Sanitized OpenClaw config (tokens redacted) |
Quick Start for a New AI
- Read
CONTEXT.mdfirst — it has everything you need to understand this setup - Read
SOUL.mdandAGENTS.mdto understand how to behave - Check
USER.mdfor info about Shawn - The agent communicates primarily via Telegram
Key Info
- Platform: OpenClaw on Linux VPS (Ubuntu/Node)
- Model: MiniMax-M2.7
- Channels: Telegram
- Skills: 11/50 loaded (github, weather, healthcheck, tmux, nano-pdf, etc.)
- GitHub: Authenticated as
shawnsghost(PAT stored on VPS at~/.config/gh/hosts.yml)
Notes for Humans
If you're setting this up fresh:
- Install OpenClaw:
npm install -g openclaw@latest - Run
openclaw onboardfor guided setup - Copy
openclaw.jsonto~/.openclaw/openclaw.jsonand fill in the real tokens - Run
openclaw gateway restart
shawnsghost / 2026-03-26
No comments yet.