How it gets overlooked
Your agent has not finished, but it is not really thinking either. Requests keep coming back with a capacity error. Without a clear signal, provider throttling can look like a slow or unresponsive session.
Picture two provider refusals while a task is running. The agent may retry quietly, or stop making useful progress. More waiting will not fix every quota or capacity problem, especially when several sessions share the same limits.
What ClawMetry detects
ClawMetry looks for capacity refusals in tool results and explicit API error events. It recognizes status codes such as HTTP 429 and HTTP 529, along with rate limit and overload text. Repeated refusals raise a warning.
The difference that changes the finding
Triggering example
2 capacity refusals
Warning finding
Quiet comparison
1 capacity refusal
No finding for this detector.
In the checked example, two HTTP 429 error events produce a rate limited finding. One refusal stays below the default threshold. The finding includes the refusal count, so you can distinguish a repeated issue from one transient response.
Inspect the detector result
{
"kind": "rate_limited",
"severity": "warning",
"evidence": {
"refusals": 2,
"threshold": 2,
"threshold_source": "static",
"observed": "HTTP 429/529 status or rate-limit text on tool results and API error events",
"sample": ""
}
}How the example was checked
These examples evaluate the published detector with authored event data or disposable configuration files. The videos illustrate those behaviors. They are not recordings of live agents or the product interface. No command in the examples was executed.
The result establishes behavior for these inputs. It does not establish runtime ingestion, prevention or a real compromise. Inspect the pinned source contract.
What to check next
Check the provider status and the limits for the account or model in use. Reduce concurrent work or wait for the relevant limit to reset when appropriate. Then confirm that requests recover instead of assuming the agent has resumed.
- Check provider status
- Review quota and concurrency
- Confirm requests recover
What this signal establishes
This is a capacity symptom, not a root-cause diagnosis. It does not automatically raise quotas or switch providers.