Abstract
Agents are now used widely in the process of software development, butbuilding production-ready software engineering agents is a complex task.Deploying software agents effectively requires flexibility in implementationand experimentation, reliable and secure execution, and interfaces for users tointeract with agents. In this paper, we present the OpenHands Software AgentSDK, a toolkit for implementing software development agents that satisfy thesedesiderata. This toolkit is a complete architectural redesign of the agentcomponents of the popular OpenHands framework for software development agents,which has 64k+ GitHub stars. To achieve flexibility, we design a simpleinterface for implementing agents that requires only a few lines of code in thedefault case, but is easily extensible to more complex, full-featured agentswith features such as custom tools, memory management, and more. For securityand reliability, it delivers seamless local-to-remote execution portability,integrated REST/WebSocket services. For interaction with human users, it canconnect directly to a variety of interfaces, such as visual workspaces (VSCode, VNC, browser), command-line interfaces, and APIs. Compared with existingSDKs from OpenAI, Claude, and Google, OpenHands uniquely integrates nativesandboxed execution, lifecycle control, model-agnostic multi-LLM routing, andbuilt-in security analysis. Empirical results on SWE-Bench Verified and GAIAbenchmarks demonstrate strong performance. Put together, these elements allowthe OpenHands Software Agent SDK to provide a practical foundation forprototyping, unlocking new classes of custom applications, and reliablydeploying agents at scale.