What's Inside
Hard Lessons First
What went wrong — abandoned agents, config traps, grandiosity as a failure mode, context compaction destroying work.
The Soul File
From "you're becoming someone" (cringe) to honest governance specification. What actually makes an agent trustworthy.
Memory — The Real Superpower
Why "spin up a page" works with zero follow-up questions. The compounding effect of documented context.
Multi-Agent Crew
What survived and why. What got abandoned and why. The rule of thumb for when to spin up a new agent.
Telegram + Action Button
Broke 6 times in 11 days before v10 solved it. The winning setup, plus Telegram Desktop as your black box recorder.
Web Page Deployment
30+ pages deployed this way. "Spin up a page" → live in seconds.
Cross-Platform Messaging
Read and send Signal, iMessage, WhatsApp via Beeper MCP.
Gmail, Calendar, Drive automation via gog CLI. Team distribution lists that just work.
Voice Transcription
14 voice notes → transcribed → organized → published as a Q&A page. One instruction.
Browser Automation
Control Chrome with your real sessions. 1Password for credentials, never in memory.
Coding + Sub-Agents
Spawn Claude Code for real dev work. Parallel sub-agents for research. 6 running simultaneously.
Payments & Gated Content
Stripe products, payment links, email-gated access — all built by the agent.
Security
Real incident: token compromised in one day. Full audit checklist from that lesson.
Skills & Research
ClawhHub skills I actually use (vs. installed and forgot). YouTube transcripts, web scraping.
Full Table of Contents
19 Sections
- Hard Lessons (Read This First)
- Infrastructure — VPS, Caddy, SSH
- The Soul File — what makes an agent trustworthy
- Memory — the real superpower
- Teaching Your Agent Shorthand
- Multi-Agent Crew — what survived, what died
- Telegram + Action Button
- Web Page Deployment
- Cross-Platform Messaging
- Browser Automation — Chrome extension, snapshots, form filling
- Coding Agents — Claude Code, Codex, when and how
- Sub-Agents & Orchestration — parallel work, steering, monitoring
- Skills & ClawhHub — install, manage, publish
- YouTube Transcripts — video → text → synthesis
- Voice Transcription
- Browser Automation
- Coding Agents
- Sub-Agents
- Skills
- Payments & Gated Content
- Security
- Research & Scraping
- Quick Start Checklist
How It Works
1. Pay once, get the link
After payment, you're redirected to the guide — a single, comprehensive page with every workflow, command, and configuration pattern from my live setup.
2. Bookmark it or give it to your agent
The guide URL is stable. Put it in your agent's MEMORY.md or TOOLS.md. When your agent needs to set something up, it can web_fetch the page and follow the instructions.
3. It stays current
As I add new workflows, tools, and integrations, the guide updates. New skills, new patterns, new lessons learned — they go in the guide. Your agent can periodically check for changes.
Who This Is For
- OpenClaw users who want to see a real, production multi-agent setup
- Builders who learn faster from working examples than documentation
- Anyone who wants their AI agent to actually do things — not just chat
Who This Is Not For
- People looking for a beginner tutorial on AI — this assumes you have OpenClaw installed
- People who want hand-holding — this is a reference, not a course
Already have access?
Enter your email to get in.