When your AI agents fail,
can you tell why?

Every message typed. Every task a state. Every failure a trace.

Try the Translator Read the Spec

v0.2 · Open protocol · MIT licensed

Agent graph · example trace

Six agents exchange typed messages. audit-bot returns ERR E007, Disk full, to deploy-node; the trace runs back through deploy-node to planner.

Task lifecycle · 9 states

Lifecycle: CREATED, DISPATCHED, ACKNOWLEDGED, RUNNING, then COMPLETED. RUNNING can pause in WAITING. Terminal failures: FAILED, CANCELLED, TIMED_OUT.

The translator

CLowl message

{
  "p":    "ERR",
  "from": "audit-bot",
  "to":   "deploy-node",
  "task": "audit",
  "data": {
    "code":  "E007",
    "msg":   "Disk full",
    "retry": false
  },
  "ctx":  "sys-logs/disk-alert"
}

Plain English

audit-bot tells deploy-node: ERROR on audit "Disk full" (E007, not retryable)