OpenClaw Agent Orchestrator

Create and repair durable OpenClaw agents, bindings, and runtime state. Use when the user wants real subagents, durable agent entries, or proof that orchestration actions actually changed live state. Prefer verified `openclaw agents add` flows over optimistic `sessions_spawn` narration.

Use this skill when the user wants real subagents or when the runtime claims it started agents but openclaw agents list and openclaw agents bindings do not show them.

Durable Agent Rule

For durable subagents in this deployment, use openclaw agents add and verify the result.

Do not use sessions_spawn with mode="session" for durable agent creation here. This runtime has already returned errors such as:

• mode="session" requires thread=true
• thread=true is unavailable because no channel plugin registered subagent_spawning hooks

Treat sessions_spawn as ephemeral-only unless the current runtime explicitly proves thread-bound session support.

Workflow

1. Run scripts/verify-runtime.sh to see the current truth.
2. If the requested durable agent does not exist, run scripts/create-agent.sh <agent-id> <name> <emoji> [theme].
3. If the user asked for routing, add bindings with openclaw agents bind.
4. Re-run scripts/verify-runtime.sh.
5. Report only what is present in verified runtime output.

Guardrails

• Never say an agent was created until openclaw agents list --json shows it.
• Never claim bindings exist until openclaw agents bindings --json shows them.
• Do not use crontab in this environment. Use openclaw cron ... only if a schedule is actually needed.
• Do not say GO, starting now, or spawned successfully before a verified state change.

Commands

Verify:

./scripts/verify-runtime.sh

Create one durable agent:

./scripts/create-agent.sh oracle "Oracle" "🔮" "qa"

Create one durable agent from a copied workspace and verify:

./scripts/create-agent.sh trinity "Trinity" "🥷" "ops"
./scripts/verify-runtime.sh