A FastAPI based python debugger

This is a very basic debugger that runs inside of a FastAPI server.

NOTE: This server is only semi-functional. It currently pauses on every line whether there is a break point or not. It also doesn't handle exceptions in the target script very well.

It's a proof of concept for a debugger that an LLM could use to debug a python script. LLMs can use OpenAPI compliant APIs easily via LangChain OpenAPI function chains!

1. Start the server

   poetry run server
   

2. Init the session

   Init the debugging with a POST to /debug/init The payload should be a classpath, where the last element is the entry function. For example, to start start() in foo.py send this payload

   {classpath: "foo.start"}
   

3. Start the script

   Start the debugging run with a POST to /debug/start

4. List variables from paused frame

   Fetch variables from the current frame with a GET to /debug/vars

5. Step to the next line

   Step to the next line with a POST to /debug/step