Start Here - OpenClaw Academy
Start Here — OpenClaw Academy
Orientation post · Always free · Read this once, then go to Issue 01
OpenClaw Academy is a practitioner-first curriculum covering AI agents and the OpenClaw platform — from the very first concept to autonomous systems running in production. Two posts per week. Working code in every paid post.
This post is your orientation. Read it once. Then go to Issue 01.
How the newsletter is structured
Concept posts — every Tuesday, always free. One idea, fully explained. An engineering problem OpenClaw reveals. The architecture behind it. Three things to carry forward. One action that takes less than ten minutes. Written to be forwarded.
Deep dives — every Thursday, free preview + paid full. End-to-end implementation. Architecture diagram or code walkthrough. Full vault asset (GitHub folder with working code, annotated README, version stamp). Free subscribers get the concept and architecture. Paid subscribers get the full implementation, vault asset, and the career signal at the end.
Every post teaches exactly one concept and leaves you with exactly one action. If you finish a post and feel like you learned five things, something went wrong. The goal is one concept, fully understood, with something running on your machine by the end.
Where to start based on your role
You are. Start here.
Developer, new to agents. Issue 01 — What Is an AI Agent →
Developer, familiar with agents Issue 03 — Context Engineering →
QA engineer. Issue 01, then Part 1 Issue 07 — Agent Evaluation
SRE / DevOps Issue 01, then Part 4 — Production Operations
Engineering manager Issue 01 — the concept posts are written for you too
Data engineer Issue 01, then Part 2 Issue 13 — RAG and Knowledge Agents
Security engineer Issue 01, then Part 1 Issues 05–06 — The Security Model
Architect / Staff engineer Any post — every deep dive has a full architecture section
The 5-part curriculum
Part 1 — Foundations (you are here) What agents are. How OpenClaw works as a distributed system. Context engineering. Skill design for production reliability. The security model. Agent evaluation. By the end: a working, secured, evaluated agent you built yourself.
Issues 01–08 · Tuesdays and Thursdays
Part 2 — Core Patterns MCP and tool integration. Memory architecture across three horizons. RAG and knowledge agents. TaskFlow and durable workflows. Multi-agent design. Agentic design patterns in running OpenClaw code.
Issues 09–16 · Tuesdays and Thursdays
Part 3 — Interfaces and Applications Channel configuration. Voice and multimodal. The morning briefing agent done correctly. Inbox triage. GitHub workflow automation. Building a complete personal operations agent that runs 24/7.
Issues 17–24 · Tuesdays and Thursdays
Part 4 — Production VPS deployment. Observability (Prometheus, Grafana). Cost engineering as a first-class discipline — right model for right task, caching strategies, token-efficient skill design. Incident response for agents. The Agent Reliability Engineer role.
Issues 25–32 · Tuesdays and Thursdays
Part 5 — Frontier Autonomous systems. Self-improving agents. The production readiness gap in OpenClaw — honest assessment of what enterprises need that OpenClaw doesn’t yet provide. NemoClaw and enterprise forks. The five-question mental model that transfers to any agentic framework.
Issues 33–40 · Tuesdays and Thursdays
The code
Every deep dive has a companion folder in the private GitHub repository:
openclaw-academy-labs/
skills-vault/
part-1/
issue-02-subsystem-map/
openclaw.json ← annotated, every field labelled by subsystem
gateway-test.sh ← verifies all 5 subsystems, exits non-zero on failure
subsystem-diagram.md
README.md ← Problem / Decision / Files / Prerequisites / Version
issue-04-skill-reliability-kit/
issue-06-hardened-config/
issue-08-eval-harness/
...
Free subscribers: the concept is in the Tuesday post. Paid subscribers: clone the folder and it runs. Every folder is self-contained — you don’t need to have read previous issues to use an individual vault asset.
The four ways to access content
Free newsletter — subscribe on Substack. Every Tuesday and Thursday concept post, free forever.
Paid subscription — $15/month or $99/year — full deep dives + GitHub vault access. One vault asset every Thursday. The vault grows with every issue.
Issue Vault — $299 one-time — all published walkthroughs collected as clean PDFs with drill cards, in a Google Drive folder. Lifetime access, no subscription required. For engineers who want everything at once or want to share with their team. Paid subscribers get it for $249.
Async Production Program — $999 per cohort — a structured 6-part course. Six parts unlock weekly via email. Written feedback on assignments within 72 hours. Private Discord cohort channel. 20 seats per cohort. No live sessions — designed for engineers with full-time jobs. Issue 01 through Part 4 assumed; this is the intensive structured path.
All four options are at vault.systemdrd.com/openclaw-academy.
Start reading
Issue 01 — What Is an AI Agent →
Every reader, every role, starts here.
Questions? Reply to any email. I read all of them. Subscribe free · Go paid · vault.systemdrd.com/openclaw-academy