a6f0c066b5
* add outline of agent * add plan class * add initial prompt * plumb plan through a bit * refactor state management * move task into state * fix errors * add prompt parsing * add task actions * better serialization * more serialization hacks * fix fn * fix recursion error * refine prompt * better description of run * update prompt * tighter planning mechanism * prompt tweaks * fix merge * fix lint issues * add error handling for tasks * add graphic for plans * remove base_path from file actions * rename subtask to task * better planning * prompt updates for verification * remove verify field * ruff * mypy * fix actions
CodeAct-based Agent Framework
This folder implements the CodeAct idea that relies on LLM to autonomously perform actions in a Bash shell. It requires more from the LLM itself: LLM needs to be capable enough to do all the stuff autonomously, instead of stuck in an infinite loop.
A minimalistic example can be found at research/codeact/examples/run_flask_server_with_bash.py:
mkdir workspace
PYTHONPATH=`pwd`:$PYTHONPATH python3 opendevin/main.py -d ./workspace -c CodeActAgent -t "Please write a flask app that returns 'Hello, World\!' at the root URL, then start the app on port 5000. python3 has already been installed for you."
Example: prompts gpt-4-0125-preview to write a flask server, install flask library, and start the server.
Most of the things are working as expected, except at the end, the model did not follow the instruction to stop the interaction by outputting <execute> exit </execute> as instructed.
TODO: This should be fixable by either (1) including a complete in-context example like this, OR (2) collect some interaction data like this and fine-tune a model (like this, a more complex route).