課程簡介

OpenClaw Foundations and Safety Model

  • What OpenClaw is, what it is not, and when it is a good fit
  • Core concepts: agent, tools, skills, memory, connectors, approvals
  • Corporate realities: data sensitivity, environment separation, and safe defaults

Setup, Configuration, and First Agent Run

  • Prerequisites check: Node.js, Git, API keys, workspace folders
  • Install OpenClaw, verify the installation, and understand the project layout
  • Connect an LLM provider, set core configuration, and validate connectivity
  • Run a starter agent with read-only actions first, then add controlled write actions

Using Built-in Tools and Reliable Prompting

  • Working with common tools: files, shell commands, and simple web tasks
  • Prompting patterns for predictable execution: constraints, step plans, and confirmations
  • Reviewing agent outputs, tool calls, and traces to catch issues early

Skills and Memory in Practice

  • Adding and configuring skills for repeatable workflows
  • Memory basics: what should be stored, what should not, and how to reset safely
  • Practical exercise: build a small workflow that uses memory carefully (with a clear stop condition)

Building and Testing a Custom Skill

  • Skill structure, inputs and outputs, and how OpenClaw discovers and runs skills
  • Implement a small business-oriented skill (example: summarize a folder of reports and produce a short brief)
  • Testing approach: sample inputs, expected outputs, error handling, and documentation

Integrations, Operations, and Next Steps

  • Integration patterns: chat and ticket workflows in a safe sandbox environment
  • Designing a repeatable automation flow: trigger, action, review, approvals, and handoff
  • Operational basics: logging, auditability, configuration management, and a pilot readiness checklist

最低要求

  • Comfort with basic command line usage (folders, paths, environment variables)
  • Ability to install and run developer tools on your workstation (Git, Node.js)
  • Basic JavaScript or scripting experience (reading and making small edits)

Audience

  • Developers and automation engineers who want to build AI-powered assistants and internal tooling
  • IT and operations professionals who want to automate recurring support and admin tasks
  • Technical product owners and team leads evaluating self-hosted AI agent options
 7 小時

人數


每位參與者的報價

即將到來的課程

課程分類