/agents-workshop-for-dev-sec-ops

Primary LanguageJupyter Notebook

Learning about Agents - With some Dev,Sec,Ops examples

Outline

We'll be asking agents powered by LLMs to create bash script.

  • We start with a simple LLM call using langchain
  • Then we show the ReAct pattern using langchain
  • After that we create a Solo Agent with CrewAI a framework for Agents
  • Next up the Ops crew joins the Dev crew to improve the script by bringing a tool that can check Bash syntax
  • Finally we show how we can add the Sec crew into a predictable workflow using Langgraph

Learnings

  • We'll demystify that it's all about prompts
  • Well show what goes over the wire to the LLM

Pre-Reqs

  • The notebooks assume you are using OpenAI

  • and it needs the ENV VAR OPENAI_API_KEY set. (create a .env file for example when using VSCode )

  • The notebooks assume you are using OpenAI

  • and it needs the ENV VAR OPENAI_API_KEY set. (create a .env file for example when using VSCode )

GitHub CodeSpaces

To keep things as simple as possible you can run this workshop in GitHub CodeSpaces. Simply hit the Launch Lab link below and a new codespace will open. The CodeSpace will be an embedded VSCode with Jupyter Notebook pre-configured and ready to go.

Note: It may take a few moments for Python to be ready. Once the Lab is ready open the 0_welcome.ipynb file to begin the lab. Open in GitHub Codespaces

Once you have finished the workshop you should Manage your codespaces and delete the codespace instance.