Metrics

Context Layer exposes per-project runtime metrics through the console Observability page. Metrics are scoped to your project.

10-Second overview

The Observability page shows daily counters and histograms for your project's runtime activity. Select a date to view metrics for that day. Metrics are available on paid plans only.

What metrics show

Counters record discrete runtime events. Histograms record distributions across latency, token usage, and prompt size.

Production plan metrics

Available on the Production plan:

Counters:

invocations_total — total runtime executions for the selected date

Histograms:

request_latency_ms — end-to-end request latency
provider_latency_ms — provider invocation latency
token_usage_prompt — prompt token consumption
token_usage_completion — completion token consumption
prompt_size_chars — prompt size in characters

Scale plan metrics

Available on the Scale plan. Includes everything in Production plus:

Counters:

tier_decisions_total — total enforcement tier decisions
constraint_violations_total — total constraint violations detected
verification_failures_total — total verification failures
replay_hits_total — total deterministic replay cache hits
provider_failures_total — total provider invocation failures

Plan access

Metrics are not available on the Free plan. The Production plan includes basic per-project metrics. The Scale plan includes full per-project metrics.

Accessing metrics

Metrics are accessible through the Context Layer console under Observability. Select a date using the date picker to view metrics for that day. Counters and histograms are displayed in separate sections.

Non-claims

Metrics reflect runtime execution behavior scoped to your project. They do not expose internal evaluation logic, decision reasoning, or pipeline mechanics.