Runtime for OpenClaw
OpenClaw is where Spinup started. Run OpenClaw in its own cloud environment today: each agent gets an isolated Firecracker microVM with projected secrets, managed configuration, persistent state, and durable run records. Hermes runs on the same model alongside it. Claude Code, OpenCode, Codex CLI, and other harnesses are on the roadmap.
'%20class='claw-body'%3e%3c/path%3e%3c!--%20Left%20Claw%20--%3e%3cpath%20d='M20%2045%20C5%2040%200%2050%205%2060%20C10%2070%2020%2065%2025%2055%20C28%2048%2025%2045%2020%2045Z'%20fill='url(%23lobster-gradient)'%20class='claw-left'%3e%3c/path%3e%3c!--%20Right%20Claw%20--%3e%3cpath%20d='M100%2045%20C115%2040%20120%2050%20115%2060%20C110%2070%20100%2065%2095%2055%20C92%2048%2095%2045%20100%2045Z'%20fill='url(%23lobster-gradient)'%20class='claw-right'%3e%3c/path%3e%3c!--%20Antenna%20--%3e%3cpath%20d='M45%2015%20Q35%205%2030%208'%20stroke='%23ff4d4d'%20stroke-width='2'%20stroke-linecap='round'%20class='antenna'%3e%3c/path%3e%3cpath%20d='M75%2015%20Q85%205%2090%208'%20stroke='%23ff4d4d'%20stroke-width='2'%20stroke-linecap='round'%20class='antenna'%3e%3c/path%3e%3c!--%20Eyes%20--%3e%3ccircle%20cx='45'%20cy='35'%20r='6'%20fill='%23050810'%20class='eye'%3e%3c/circle%3e%3ccircle%20cx='75'%20cy='35'%20r='6'%20fill='%23050810'%20class='eye'%3e%3c/circle%3e%3ccircle%20cx='46'%20cy='34'%20r='2'%20fill='%2300e5cc'%20class='eye-glow'%3e%3c/circle%3e%3ccircle%20cx='76'%20cy='34'%20r='2'%20fill='%2300e5cc'%20class='eye-glow'%3e%3c/circle%3e%3cdefs%3e%3clinearGradient%20id='lobster-gradient'%20x1='0%25'%20y1='0%25'%20x2='100%25'%20y2='100%25'%3e%3cstop%20offset='0%25'%20stop-color='%23ff4d4d'%3e%3c/stop%3e%3cstop%20offset='100%25'%20stop-color='%23991b1b'%3e%3c/stop%3e%3c/linearGradient%3e%3c/defs%3e%3c/svg%3e)
What Makes OpenClaw Different
A full agent runtime, not a thin model wrapper
OpenClaw has its own state layer, configuration system, and resource requirements. A generic sandbox overlooks most of those details. The runtime needs to understand how OpenClaw actually works.
Managed installation and runs
Each run is a self-contained CLI call. The runtime handles installation, dependency management, and process lifecycle so your team never touches Node toolchains.
Built-in session continuity
Multi-turn conversations maintain context across runs. The agent's state persists between runs, so it picks up where it left off instead of replaying from scratch.
Right-sized resources per harness
OpenClaw's runtime stack (Node, SQLite state layer, and the agent process) needs more headroom than lighter harnesses. Spinup allocates the right resources per harness automatically.
Zero-setup configuration
Spinup projects all required configuration into the environment automatically. Config validation runs before the first run, catching misconfigurations early rather than mid-run.
Reference Implementation
How OpenClaw shaped the Spinup runtime
Every runtime makes design decisions based on the first real workload it supports. For Spinup, that workload was OpenClaw. The need for persistent state shaped the environment model. The need for secure credentials led to secrets projection. The need to catch broken configs early led to harness lifecycle hooks.
OpenClaw runs on Spinup today through `/exec` inside an isolated microVM, with managed secrets and persistent state. The runtime handles what the harness does not: isolation boundaries, environment durability, and operational controls that work across every harness on the platform.
Because OpenClaw came first, every harness that follows inherits the same primitives. The team did not design abstractions in theory and then apply them. They extracted them from making OpenClaw work in production, then generalized.
FAQ
Common questions about running OpenClaw on Spinup
What does it mean that OpenClaw is Spinup's reference harness?+
OpenClaw was the first harness Spinup integrated, and it's one of the two supported harnesses today, alongside Hermes. The runtime primitives (microVM isolation, secrets projection, durable run records, harness lifecycle) were shaped against OpenClaw's real requirements before any other harness existed. That makes it the reference implementation, not just one option in a list.
Why does OpenClaw need more resources than other harnesses?+
OpenClaw's runtime stack (Node, SQLite state layer, and the agent process) needs more headroom than lighter harnesses. Spinup allocates resources at the harness level, so OpenClaw gets the memory it requires without starving other agents on the same account.
How does session continuity work for OpenClaw on Spinup?+
OpenClaw supports single-turn and multi-turn runs. Spinup runs it through `/exec` inside the agent's isolated microVM and preserves the environment state between runs, so multi-turn sessions pick up where they left off instead of replaying the full history.
Do I need to configure OpenClaw manually inside the environment?+
Spinup projects all required configuration into the environment automatically. You do not need to set up config paths, state directories, or runtime flags by hand. The runtime validates configuration before the first run to catch issues early.
Related
OpenClaw fits inside a broader runtime model
The runtime stays stable above the harness layer. These pages cover the primitives that OpenClaw, and every other harness, depends on.
Early access
Run OpenClaw in its own cloud environment. Keep the harness swappable.
Request access if this is the runtime shape your team has been missing.