OpenHands

An autonomous software engineering agent that works inside a sandbox, reading a repository, editing files and running commands to finish assigned tasks without a person in the loop. It picks up issues, implements changes, runs tests and iterates on failures, through a terminal or a browser interface with a code editor attached.

Teams use it for scoped maintenance work, issue triage, test writing and repetitive refactors, and developers use it to explore unfamiliar codebases.

Running it on infrastructure you own keeps source code, prompts and command output inside your perimeter, avoids paying a hosted agent per task, and lets you reset the container between runs. It can be hosted on your own infrastructure.