OpenHands launches Agent Canvas control center

Visual control planes like Agent Canvas are the logical next step for AI engineering, bridging the gap between isolated CLI chats and production-ready, multi-agent workflows.

• –ACP Integration: Leverages the Agent Client Protocol to decouple the UI from agent runtimes, allowing developers to switch between various backend agents like Claude Code and Codex seamlessly.

-

HEADLINE: OpenHands has launched Agent Canvas, a self-hosted control center that orchestrates and automates multiple coding agents across local, VM, and cloud backends.

PRODUCT_NAME: OpenHands Agent Canvas

SUMMARY: OpenHands Agent Canvas is an open-source, self-hosted developer control center designed to manage and automate multiple AI coding agents. By supporting agents like OpenHands, Claude Code, and Codex through the Agent Client Protocol (ACP), it allows developers to switch between local, Docker, VM, and cloud environments without losing context. Additionally, users can create event-driven or scheduled workflows integrating Slack, GitHub, and Linear to build an always-on automated engineering team.

ANALYSIS: Orchestrating multiple specialized agents is the future of AI engineering, and visual control centers like Agent Canvas are crucial for making these workflows manageable.

• –Decoupled Orchestration: Standardizing communication via the Agent Client Protocol (ACP) allows new and existing agents to be integrated without UI redesigns.
• –Infrastructure Flexibility: Seamless switching between local development, isolated Docker containers, VMs, and cloud backends handles resource-intensive tasks without local hardware limitations.
• –Operational Automations: Moving beyond simple one-off prompts to triggerable event-driven workflows (via Slack, GitHub, Linear) unlocks genuine engineering autonomy.