AIFor agents: /use-cases/see-everything.llms.txt

See everything

When something degrades at 3 a.m., "the RPC feels slow" isn't a diagnosis. eRPC measures every request at every hop — which project, which chain, which upstream, which method, cache hit or miss, retried or hedged — and exposes it all as Prometheus metrics and OpenTelemetry traces. Grafana dashboards ship in the repo. Your on-call sees exactly which provider degraded and when, instead of guessing.

Monitoring & metricsThe 20 metrics that matter on day one, plus ready-made dashboards.Tracing & loggingFollow one request through every policy and upstream attempt.HealthcheckReadiness that tells your load balancer the truth.Error taxonomyEvery error normalized to a named code you can alert on.Metrics referenceAll 122 metrics, every label, in one table.

All of the above in one place — illustrative, not a tuned production config:

(root)

erpc.yaml

# structured logs with secret redactionlogLevel: info# Prometheus on :4001/metricsmetrics:  enabled: true  port: 4001# one OTel trace per request, every hop; keep 10% of tracestracing:  enabled: true  protocol: grpc  endpoint: otel-collector:4317  sampleRate: 0.1# /healthcheck is always on; error codes are normalized automatically.

Agent reference

Copy one of these prompts into your AI agent session (Claude Code, Cursor, …) — each one points the agent at this page's machine-readable reference so it can do the work correctly:

Prompt Example #1: wire up Prometheus metrics and a Grafana dashboard

I want to enable eRPC metrics so I can import the bundled Grafana dashboard and
start monitoring upstream health and cache hit rates. Update my eRPC config with the
metrics config and tell me which metrics matter most on day one. Read the full
reference first: https://docs.erpc.cloud/use-cases/see-everything.llms.txt

Prompt Example #2: add distributed tracing and structured logging

I need to trace individual requests through eRPC all the way to the upstream to
debug slow responses. Configure OpenTelemetry tracing and structured log output in
my eRPC config with a sensible sample rate for production. Reference:
https://docs.erpc.cloud/use-cases/see-everything.llms.txt

Prompt Example #3: build alerts for upstream degradation and chain-tip lag

I want Prometheus alerts that fire when an upstream starts erroring heavily or falls
behind the chain tip. Walk me through the right metric names, labels, and threshold
values to use, given my eRPC setup in my eRPC config. Reference:
https://docs.erpc.cloud/use-cases/see-everything.llms.txt

See everything — agent starting pointsExpand for every option, default, and edge case — or copy this entire section into your AI assistant.

This page sells the outcome; implementation lives in the feature pages. Fetch their machine-readable companions:

Monitoring & metrics — metrics server config, key operational metrics, Grafana dashboards.
Tracing & logging — OTel endpoint/sampling/protocol, span hierarchy, log levels, secret redaction.
Healthcheck — eval strategies, root vs per-project checks, k8s probes.
Error taxonomy — every error code, retryability, HTTP status mapping.
Metrics reference — the complete 122-metric catalog with labels and cardinality warnings.

Alerting hints for agents: upstream health is erpc_upstream_request_errors_total by error/severity, chain-tip freshness is erpc_network_latest_block_timestamp_distance_seconds, and benching events surface via cordon metrics — exact names, labels, and firing conditions are in the metrics reference agent section.

Lock it down Docker