Advanced · Free course
OpenClaw: Build Your Own AI Agent
How an AI agent harness works, file by file, and how to build your own.
60 lessons0 videos so farNotes for every lesson
What you’ll learn
- The agent loop
- Tools and memory
- The OpenClaw harness
- Your personal agent
0 of 60 complete
Curriculum
- 1What an AI agent is, and what OpenClaw issoon
- 2The big picture: OpenClaw's architecturesoon
- 3Installing OpenClawsoon
- 4Model providers and API keys, safelysoon
- 5Onboarding and your first conversationsoon
- 6The agent workspacesoon
- 7Configurationsoon
- 8Models, thinking and failover basicssoon
- 9What a harness issoon
- 10The agent loop I: one turnsoon
- 11The agent loop II: tool callssoon
- 12Streamingsoon
- 13Tool batches and parallel toolssoon
- 14When the model gets tools wrong: recovery and repairsoon
- 15Guards against runaway agentssoon
- 16Interruption and steeringsoon
- 17The system prompt, assembledsoon
- 18Context and the context enginesoon
- 19Compaction I: when and where to cutsoon
- 20Compaction II: summarisingsoon
- 21Sessions and transcriptssoon
- 22The session controller: one writer at a timesoon
- 23A run from start to finishsoon
- 24Errors, retries and model failoversoon
- 25Shaping the replysoon
- 26Hooks I: internal hookssoon
- 27Hooks II: plugin hookssoon
- 28Reasoning and thinkingsoon
- 29Agent runtimessoon
- 30Plugging in your own harnesssoon
- 31Debugging a run: trajectories and logssoon
- 32Usage and cost trackingsoon
- 33Build a mini harness I: the loop and one toolsoon
- 34Build a mini harness II: streaming, parallel tools, interruptionsoon
- 35Build a mini harness III: transcript and compactionsoon
- 36Built-in toolssoon
- 37exec and exec approvalssoon
- 38Sandboxingsoon
- 39Skillssoon
- 40Write your first skillsoon
- 41MCP in OpenClawsoon
- 42Pluginssoon
- 43Memory Isoon
- 44Memory IIsoon
- 45Sub-agents and multi-agent routingsoon
- 46Secrets and the security modelsoon
- 47An agent that acts on its ownsoon
- 48The gatewaysoon
- 49Channelssoon
- 50Clients, Control UI and the HTTP APIssoon
- 51Nodes and devicessoon
- 52Running it somewhere elsesoon
- 53Watching it in productionsoon
- 54Your personal agent I: the designsoon
- 55Your personal agent II: workspace, model and memorysoon
- 56Your personal agent III: skills for daily worksoon
- 57Your personal agent IV: a custom tool plugin and hookssoon
- 58Your personal agent V: Telegram, approvals and sandboxsoon
- 59Your personal agent VI: testing and evaluationsoon
- 60Your personal agent VII: running it 24/7, safelysoon