DigitalOcean puts Managed Agents into public preview for AI workloads
The service gives agent builders a managed runtime and controlled tool access instead of making them assemble that layer themselves.
DigitalOcean has put Managed Agents into public preview, giving teams a managed layer for AI-agent workloads with isolated microVM runtimes, controlled tool access, and serverless inference. The offering is split into a Harness Runtime and an Action Gateway: the runtime keeps agent state and sessions alive across pauses, resumes, and forks, while the gateway exposes external tools through a single MCP endpoint with permission controls and human approval for risky actions. DigitalOcean says this is meant to remove the plumbing developers usually have to build themselves for persistence, coordination, secure tool use, and rapid startup of agent workloads. The supported harnesses include Claude Code, Codex CLI, OpenCode, Hermes, and custom LangGraph agents, and the gateway currently advertises more than 16,000 tools. The article also notes that, unlike Docker Cloud Sandboxes, DigitalOcean’s new product is aimed more at production infrastructure and orchestration than at moving sandboxes between local and cloud environments.
Why it matters
For teams running AI agents, the main change is that persistence and tool use move into a managed service rather than custom infrastructure. That can simplify how agent sessions survive pauses, resumes and forks, while still keeping external actions behind permission checks and human approval. It also broadens the set of agent frameworks that can connect to the same tool layer, including Claude Code, Codex CLI, OpenCode, Hermes and LangGraph agents.
Keep or strike?
Does this story matter, or is it hype? Mark it before you see what everyone else did.
Sources
- InfoQ