Kova/docs/OCM_OPERATOR_INTEGRATION.md

OCM Operator Integration

Kova uses OCM as the OpenClaw environment and runtime control plane. Agents that have access to an ocm-operator skill should load it before executing real Kova scenarios.

The repo-local OCM skill lives at .agents/skills/ocm-operator.

Why

Kova scenarios can create disposable envs, clone existing user state, build local release-shaped runtimes, upgrade envs, inspect services, collect logs, and clean up. Those operations are exactly where agents need OCM-specific safety rules.

The ocm-operator skill helps agents understand:

• envs are the isolation boundary for OPENCLAW_HOME
• durable user envs must be cloned before tests
• release behavior must use package/runtime builds, not source/dev commands
• ocm @<env> -- ... is the normal way to run OpenClaw inside an env
• service status and logs should be collected through OCM
• cleanup should use ocm env destroy <env> --yes
• local OpenClaw source should be tested through release-shaped runtime builds

Boundary

The skill is an agent operating guide. It is not a Kova runtime dependency.

Kova should remain usable by humans and CI with only:

• Node.js
• OCM on PATH
• OpenClaw runtime/network prerequisites for the selected scenario

Agent Rule

Before running a real scenario, an agent should:

1. Load/read ocm-operator if available.
2. Run node bin/kova.mjs setup --json.
3. Run node bin/kova.mjs plan --json.
4. Dry-run the intended scenario.
5. Execute exactly the intended scenario with --execute.
6. Read the JSON report.
7. Summarize only the important OpenClaw findings and cleanup status.