Summary
Add a Models tab to the AgentWatch dashboard showing per-model token usage, cost, request count, and latency.
Status
⚠️ Partially complete — infrastructure shipped in Night 7, but tab shows no data.
Done
- Models tab UI at
/models with per-model requests, tokens, cost, p50/p95 latency and cost share bars
- Storage:
model_usage table with record_model_usage() + get_model_stats()
- HTTP ingestion:
POST /api/v1/ingest/model_usage
- Public API:
agentwatch.record_model_usage()
- 10s auto-refresh
Blocked
OpenClaw gateway does not currently emit model:used (or equivalent) events, so the hook cannot capture model invocations. The index.ts handler in examples/openclaw-hook/ is a forward-looking implementation waiting for these events.
See #10 for the investigation into how to populate this tab with real data.
Summary
Add a Models tab to the AgentWatch dashboard showing per-model token usage, cost, request count, and latency.
Status
Done
/modelswith per-model requests, tokens, cost, p50/p95 latency and cost share barsmodel_usagetable withrecord_model_usage()+get_model_stats()POST /api/v1/ingest/model_usageagentwatch.record_model_usage()Blocked
OpenClaw gateway does not currently emit
model:used(or equivalent) events, so the hook cannot capture model invocations. Theindex.tshandler inexamples/openclaw-hook/is a forward-looking implementation waiting for these events.See #10 for the investigation into how to populate this tab with real data.