Skip to content

Conversation

@utsavgarg
Copy link

This PR introduces the SWE Agent, a new autonomous agent designed to solve software engineering problems from the SWE-bench and TerminalBench benchmarks.

The agent operates autonomously within a Dockerized environment, it has a extensible toolset providing basic functionality for solving software engineering tasks.

For detailed information on architecture, setup, testing, and evaluation, please refer to the README.md file included in the agent's directory.

Copy link
Member

@tpryan tpryan left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Couple nits. I am pulling in a python reviewer to look at the actual code.

Copy link
Collaborator

@happyhuman happyhuman left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

1- Please run pylint on all the python files.
2- Please choose a less generic project name other than swe_agent (which could mean anything). Also the agent names use - instead of _.

@happyhuman
Copy link
Collaborator

Thanks for all the changes. Can we rename this sample to something less generic. swe-agent is just too broad and it could mean anything (e.g. Android Developer, Web Developer, etc).

@utsavgarg
Copy link
Author

utsavgarg commented Nov 17, 2025

Thanks for all the changes. Can we rename this sample to something less generic. swe-agent is just too broad and it could mean anything (e.g. Android Developer, Web Developer, etc).

Thanks for the reviews @happyhuman, I've renamed the agent to swe-benchmark-agent as it demonstrates SWE capabilities through two popular SWE benchmarks.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants