Harness Runtime runs Codex CLI, Claude Code, OpenCode, LangGraph, and Hermes in isolated DigitalOcean sandboxes, configured with a YAML environment spec or the Control Panel and driven with doctl.
DigitalOcean Harness Runtime Conceptspublic
Generated on 24 Sep 2026
DigitalOcean Harness Runtime combines the functionality of a lightweight microVM, built-in tools like chromium and a coding sandbox needed by agents to do work. The product offers rich lifecycle APIs that persist conversational history and working state across sessions, with pause/resume/fork semantics so that developers can control costs and adapt workflows to the nonlinear quirks of agentic work. See What You Can Build for example use cases.
How Harness Runtime, Action Gateway, and the DigitalOcean services around them fit together to run and govern AI agents.
Harness Runtime adapters select which coding agent or framework runs in the sandbox and which platform features, such as approvals and checkpoints, apply to the session.
A Harness Runtime environment is a reusable configuration object defining what an agent needs to do work, and a sandbox is the isolated microVM a session provisions from it.
A Harness Runtime session is one running instance of an environment, with its own sandbox, state, and history, and a run is a single agent turn inside it.
Harness Runtime permission policies allow, require approval for, or deny the actions an agent takes, and set the filesystem and network access of the sandbox it runs in.
Harness Runtime pauses an agent and requests human approval when its permission policy marks an action as ask, and you can approve, reject, or defer from the terminal or out of band.
How Harness Runtime environments carry configuration, credentials, and scoped secrets, which bind a credential to one HTTPS destination so it never enters the sandbox.
A Harness Runtime sandbox reaches any host by default, and naming even one allowed host turns on an allowlist that denies everything else.