We (the Princeton SWE-bench team) built an agent in ~100 lines of code that does pretty well on SWE-bench, you might enjoy it too: https://github.com/SWE-agent/mini-swe-agent
replies(7):
The whole thing runs on these prompts: https://github.com/SWE-agent/mini-swe-agent/blob/7e125e5dd49...
Your task: {{task}}. Please reply
with a single shell command in
triple backticks.
To finish, the first line of the
output of the shell command must be
'COMPLETE_TASK_AND_SUBMIT_FINAL_OUTPUT'.
https://github.com/SWE-agent/mini-swe-agent/blob/7e125e5dd49...