observability-victoriametrics-tempolisted
Install: claude install-skill tstapler/stapler-squad
# Reading stapler-squad's Observability Stack (VictoriaMetrics + Tempo)
This repo's OpenTelemetry pipeline (see `docs/how-to/enable-opentelemetry.md`) fans out locally to a
docker-compose stack (`observability-*` containers) instead of only Datadog:
```
stapler-squad (OTEL_ENABLED=true)
--OTLP grpc:4317/http:4318--> otel-collector-contrib
traces --otlp exporter--> Tempo (localhost:3200, no auth)
metrics --prometheus_remote_write--> VictoriaMetrics (localhost:8428, no auth)
logs --> debug exporter only (no log backend wired up — check the app's own log file instead)
stapler-squad --profile (pprof HTTP endpoint, :6060, independent of OTEL_ENABLED)
<--pull scrape-- Grafana Alloy (alloy-config.alloy) --> Pyroscope (localhost:4040, no auth)
```
Collector config (if you need to check the pipeline itself, not just query it):
`~/dotfiles/stapler-scripts/observability/otel-collector-config.yaml`; profiling scrape config:
`~/dotfiles/stapler-scripts/observability/alloy-config.alloy`. Grafana itself runs at
`localhost:48300` (container's internal `:3000` mapped out) with VictoriaMetrics + Tempo + Pyroscope
as datasources and a provisioned `stapler-squad` dashboard folder, but for agent use, query the
backends directly — faster, no auth, no UI to screenshot. For building/editing the dashboards
themselves (not just querying live data), see the `observability-grafana-dashboards` skill.
**Before querying anything**: confirm the live service act