LangGraph: Building a Dangerously Smart AI Agent - Dynamic Code Generation & Execution!

[@smehra]

This is seriously good, so much superior to other Langgraph tutorials.

[@tharunbhaskar6795]

Very simple to follow. This clearly explains how we can make agents interact with each other and the importance of State in LangGraph. I really liked the code execution part, where we run it in isolation to protect from running potential harmful code

[@MukulTripathi]

Safety with AI generated code is needed. I'd be extending this into my voice assistant with more safety integrated into it.
We can also have a human in the loop if we want to run shell scripts on host machine.

[@tharunbhaskar6795]

@@MukulTripathi Yah, we can even use RegEx, to initially screen if the code generated contains any malious imports (like subprocess)

[@ronwiltgen2698]

Would it be possible to add this as a prefect flow? And then dynamically invoke it?

[@MukulTripathi]

Of course! In fact my voice assistant is going to be doing exactly this.

[@tharunbhaskar6795]

@@MukulTripathi Can we have a video on this

[@MukulTripathi]

For sure, I'm on an international trip for about a month, but I might find time for this. Otherwise, it'll be around end of November.

[@tharunbhaskar6795]

@@MukulTripathi Sure, will be waiting

[@techchanx]

tried everything to get langraph and langchain installed and it failed repeatedly with various build errors. gave up.

[@MukulTripathi]

Did you do a pip install for requirements? I just set it up on a new machine and it works.
Would you like to paste the error?

[@techchanx]

@@MukulTripathi got it working, tried it next day and it worked :)

[@MukulTripathi]

I'm glad it worked for you :)