OpenHands

What it is

OpenHands is an open agent framework where the LLM browses files, writes code, runs commands and tests in a sandboxed VM — like a junior engineer that works tirelessly while you sleep.

How Vaaani uses it

• Automating repetitive code refactors across large repos
• Triaging GitHub issues and drafting first-pass PRs
• Building internal devtools that act on a real shell
• Research into agent benchmarks and trajectory evaluation

Why it makes the cut

When the AI worker is itself a developer, OpenHands gives me a battle-tested harness instead of building agent infrastructure from scratch.

Sample code

from openhands.core.config import AppConfig
from openhands.runtime import get_runtime_cls

config = AppConfig(
    llm="claude-opus-4-7",
    workspace="/tmp/agent")
runtime = get_runtime_cls(config)()
runtime.execute("refactor: extract email validator")