Monitor Tool Errors

Tool reliability is often the first OpenClaw signal worth monitoring.

Recommended events:

• emit tool.calls.count for every tool invocation
• emit tool.errors.count for every failed invocation

Starter policy:

• tool.errors.count with sum > 5 in hour

Useful chart ideas:

• tool errors over time
• errors by meta.tool_name
• calls versus errors to spot failure-rate spikes

Example payload:

{
  "agent": "openclaw.main",
  "event": "tool.errors.count",
  "value": 1,
  "meta": {
    "tool_name": "browser.open",
    "workflow": "triage"
  }
}