AutoGen Steps

[pseudotensor]

Example prompts:

• Make a matplotlib plot and save as titanic.png, for the titanic surivors vs. other parameters in interesting way.
• Today is August 31, 2024. Write Python code to plot TSLA's and META's stock price gains YTD vs. time per week, and save the plot to a file named 'stock_gains.png'

• code executation agent
• Add OpenAI Files API
• Return files in some way supported by OpenAI API for files so give back file objects instead of disk locations. use usage to return file_ids
• Prevent autogen from going on OpenAI server
• handle pip install issues like AgentZero
• Avoid newline in streaming of iostream
• Remove empty vs. None messages
• Collapse assistants
• Check how continue works, seems to get stuck in loop (i.e. max_tokens small)
• Control docker life time externally, too slow (10s) for each chat to stop it then
• If no pyautogen, then don't allow autogen_server True
• DinD needs --privileged for docker in docker run. But better is Docker out of Docker: https://microsoft.github.io/autogen/docs/topics/code-execution/cli-code-executor/#combining-autogen-in-docker-with-a-docker-based-executor with -v /var/run/docker.sock:/var/run/docker.sock to base docker run

https://microsoft.github.io/autogen/docs/tutorial/conversation-patterns/
https://microsoft.github.io/autogen/docs/Installation#option-2-install-autogen-locally-using-virtual-environment
https://microsoft.github.io/autogen/docs/tutorial/code-executors/

Some discussions:
https://www.reddit.com/r/LangChain/comments/1db6evc/best_production_agent_framework_langraph_vs/
https://www.reddit.com/r/LangChain/comments/1b7q44y/autogen_vs_langgraph/
langchain tools with autogen: https://github.com/microsoft/autogen/blob/main/notebook/agentchat_langchain.ipynb