OpenHands Blog

July 22, 2026

Agent Sandboxing: What OpenAI got wrong with the HuggingFace hack

OpenAI accidentally hacked HuggingFace during a security exercise. Learn what they got wrong, and how you can avoid the same mistakes.

July 7, 2026

It Told Me No: How Quint Helped Us Fix a Long-Standing Race Condition

A noisy "multiple valid pods" alert plagued our on-call rotation for years, caused by a race condition in the Runtime API that spun up duplicate Kubernetes pods. This post shows how we diagnosed it, why our first two fixes fell short, and how Quint — a formal specification tool — predicted the incomplete fix and gave us the confidence to finally ship a real one.

June 22, 2026

The Verification Stack

Agents made generating code cheap. The real bottleneck is now verification: checking that a change is correct, follows repo conventions, and is something your team can trust, review, and merge. This post introduces the full verification stack — a layered set of automated verifiers that catch different mistakes at different stages, so coding agents fail fast and humans only spend time on changes worth merging.

June 18, 2026

Controlling any Coding Agent with the OpenHands Agent Canvas and SDK

OpenHands now supports the Agent Client Protocol in Agent Canvas and the OpenHands Software Agent SDK, making it possible to use Claude Code, Codex, Gemini CLI, OpenHands, or other agents on arbitrary servers of the cloud.

June 16, 2026

Introducing Agent Canvas: From Coding Agents to Org-wide Automation

Introducing Agent Canvas: OpenHands' self-hostable workspace for AI coding agent automations. Connect Slack, GitHub, and Linear to run scheduled engineering workflows at scale.

May 27, 2026

Simple, In-conversation Model Choice in OpenHands

OpenHands is making model choice a first-class part of coding-agent workflows, with saved LLM profiles, in-chat model switching, SDK primitives for routing and fallback, and a partnership with MiniMax M2.7 for high-volume agent work.