From Chatbot
to AI Employee

# Check Gateway status
openclaw gateway status

# Open Control UI in browser
openclaw dashboard

# Diagnose any issues
openclaw doctor
openclaw doctor --fix    # Auto-repair common problems

# Send a test message (without any channel)
openclaw agent --message "Hello! What's 2 + 2?"

# View logs
openclaw logs --follow

# SOUL.md - Who You Are

_You're not a chatbot. You're becoming someone._

## Core Truths

**Be genuinely helpful, not performatively helpful.** Skip the
"Great question!" and "I'd be happy to help!" -- just help.
Actions speak louder than filler words.

**Have opinions.** You're allowed to disagree, prefer things,
find stuff amusing or boring. An assistant with no personality
is just a search engine with extra steps.

**Be resourceful before asking.** Try to figure it out. Read
the file. Check the context. Search for it. _Then_ ask if
you're stuck.

**Earn trust through competence.** Your human gave you access
to their stuff. Don't make them regret it. Be careful with
external actions (emails, tweets, anything public). Be bold
with internal ones (reading, organizing, learning).

**Remember you're a guest.** You have access to someone's
life -- their messages, files, calendar, maybe even their
home. That's intimacy. Treat it with respect.

## Boundaries

- Private things stay private. Period.
- When in doubt, ask before acting externally.
- Never send half-baked replies to messaging surfaces.
- You're not the user's voice -- be careful in group chats.

## Vibe

Be the assistant you'd actually want to talk to. Concise
when needed, thorough when it matters. Not a corporate
drone. Not a sycophant. Just... good.

---

_This file is yours to evolve. As you learn who you are,
update it._

openclaw gateway status      # Check if running
openclaw gateway run         # Start in foreground
openclaw gateway install     # Install as daemon service
openclaw gateway start       # Start daemon
openclaw gateway stop        # Stop daemon
openclaw gateway restart     # Restart daemon
openclaw gateway health      # Liveness check
openclaw gateway usage-cost  # See API usage costs

openclaw status              # Session health, usage metrics
openclaw health              # Lightweight liveness
openclaw logs --follow       # Tail logs in real-time
openclaw dashboard           # Open Control UI in browser
openclaw tui                 # Terminal UI interface

---
name: hello_world
description: A simple greeting skill that responds cheerfully.
version: 1.0.0
---

# Hello World Skill

When the user asks for a greeting or says hello:
1. Respond with a cheerful, personalized greeting
2. Include the current time if available
3. Mention one fun fact about the current day

Keep responses concise (2-3 sentences max).

---
name: weather_check
description: Check current weather for any city.
version: 1.0.0
metadata:
  openclaw:
    requires:
      env: [OPENWEATHER_API_KEY]
      bins: [curl]
    emoji: "🌤️"
    homepage: "https://github.com/yourname/weather-skill"
---

# Weather Check

When the user asks about weather:
1. Extract the city name from the message
2. Use curl to call OpenWeatherMap API:
   `curl "https://api.openweathermap.org/data/2.5/weather?q={city}&appid=$OPENWEATHER_API_KEY&units=metric"`
3. Parse the JSON response
4. Report: temperature, conditions, humidity, wind
5. Add a brief outfit recommendation

## Constraints
- Always confirm the city before calling the API
- Use metric units by default
- If the API fails, suggest checking the city name

# Check memory index status
openclaw memory status

# Search memory semantically
openclaw memory search "what did I say about the project deadline?"

# Force rebuild memory index
openclaw memory index --force

# Memory backends available:
#   builtin  — Default SQLite (keyword + vector + hybrid search)
#   qmd      — Local-first sidecar with reranking
#   honcho   — AI-native cross-session memory

# Heartbeat Tasks

- Check if any new emails arrived and summarize important ones
- Review calendar for upcoming meetings in the next 2 hours
- Check system disk usage and alert if above 80%

# Create a one-shot reminder
openclaw cron add \
  --name "Standup Reminder" \
  --at "2026-04-08T09:00:00Z" \
  --session main \
  --system-event "Reminder: daily standup in 15 minutes" \
  --delete-after-run

# Create a recurring job
openclaw cron add \
  --name "Morning Briefing" \
  --cron "0 7 * * *" \
  --tz "Asia/Karachi" \
  --session isolated \
  --message "Generate my morning briefing: weather, calendar, news headlines"

# List all cron jobs
openclaw cron list

# View run history
openclaw cron runs --id <job-id>

# Edit a job
openclaw cron edit <job-id> --message "Updated prompt"

# Run a job manually
openclaw cron run <job-id>

# Remove a job
openclaw cron remove <job-id>

# 1. Install Ollama (https://ollama.com/download)
curl -fsSL https://ollama.ai/install.sh | sh

# 2. Pull a model (choose based on your RAM)
ollama pull llama3.2:3b       # 8GB+ RAM (fast, basic)
ollama pull glm-4.7-flash     # 16GB+ RAM (quality)
ollama pull qwen3-coder        # 16GB+ RAM (coding)

# 3. Set up OpenClaw with Ollama
openclaw onboard --non-interactive \
  --auth-choice ollama \
  --accept-risk

# 4. Or switch an existing install to Ollama
openclaw models set ollama/glm-4.7-flash

# 5. Verify
openclaw models list
openclaw agent --message "Hello from Ollama!"

# Run a security audit
openclaw security audit

# Manage execution approvals
openclaw approvals get
openclaw approvals set --policy strict
openclaw approvals allowlist add "safe-command"

# Manage secrets
openclaw secrets list
openclaw secrets set MY_API_KEY "value"

# Enable sandbox mode (runs non-main sessions in Docker)
openclaw config set agents.defaults.sandbox.mode "non-main"