OpenClaw as a runtime shape
OpenClaw-backed agents plug into OpenOrange through identity, channels, tools, health, requests, model routes, usage, chat context, and direct audited actions.
// operate.openclaw_runtime
OpenClaw runtime operations
OpenOrange runs OpenClaw-backed agents inside private workspaces with chat, channels, tools, health, local secrets, model policy, requests, and usage accounting.
OpenClaw is the deepest runtime integration in OpenOrange today. The instance dashboard exposes OpenClaw agents, chat, context usage, commands, channels, model policy, health, requests, usage, and controlled maintenance without making OpenClaw the final product boundary.
01
OpenClaw agents are represented as operated dashboard surfaces.
02
Agent chat includes runtime identity, context usage, queue state, and slash commands.
03
Runtime state remains with OpenClaw while OpenOrange keeps attributable product and audit context.
04
The same workspace is expanding to Hermes and IronClaw.
Dashboard visibility
Admins can work across OpenClaw agents, ownership, runtime health, heartbeats, chat context, commands, model policy, request attribution, and usage from the same private instance surface.
Not the whole boundary
The product is bigger than OpenClaw. Hermes and IronClaw integrations are on the way so teams can choose the runtime that fits each agent.
// questions
Does OpenOrange require OpenClaw?
No. OpenClaw is integrated today, while Hermes and IronClaw integrations are on the way and custom runtime boundaries remain explicit.
What should OpenClaw keep locally?
Execution details, sessions, memories, media, raw logs, local caches, and secret values should stay in the runtime boundary by default.