Agent skill · data analytics · jeremylongshore
clay-observability
Monitor Clay enrichment pipeline health, credit consumption, and data
Why this skill is useful
Provides specific metrics collection patterns and integration with Prometheus for monitoring Clay enrichment pipelines.
What it needs
Requires prom-client installed locally. Requires clay account access. About 4k tokens when loaded. Last updated 2026-08-07. 2,604 stars on the source repository.
What this skill does
Clay Observability Overview Monitor Clay data enrichment pipeline health across four dimensions: credit consumption velocity, enrichment success rates (hit rates), data quality scores, and CRM sync reliability. Clay's credit-based pricing model makes observability essential for cost control. Prerequisites Clay account with table access Metrics infrastructure (Prometheus/Grafana, Datadog, or custom) Webhook receiver that logs enrichment results Understanding of your enrichment column configuration Instructions Step 1: Instrument Your Clay Webhook Handler Step 2: Set Up Prometheus Metrics (Optional) Step 3: Configure Alerting Rules Step 4: Build a Dashboard Key panels for a Clay observability dashboard: Step 5: Daily Summary Report Error Handling Issue Cause Solution ------- ------- ---------- Credits depleting fast High waterfall depth or uncapped tables Add credit burn alert, reduce waterfall Hit rate near 0% Invalid input data (personal domains, typos) Add data quality monitoring, pre-filter Missing metrics Webhook handler not instrumented Add metrics collection to callback handler Dashboard shows stale data Metrics not being pushed Verify Prometheus scrape config Output Publish a privacy-aware operations view with aggregate throughput, latency, failure class, queue depth, credit consumption, quality signals, alert status, and runbook owner. Metrics and reports must not contain raw lead attributes, webhook secrets, CRM credentials, or unnecessary customer identifiers. Examples Configure a staging alert for rapid credit burn and validate it with a capped synthetic batch, then verify that the resulting report contains aggregate counts only. If a quality or error alert fires in production, freeze the affected automation path and use the incident runbook before altering metrics or suppressing the alert. Resources Prometheus Client Libraries Grafana Dashboard Examples Clay University -- Actions & Data Credits Next Steps For incident response, see clay-incident-runbook.
How to use it
Reference it in AdaL, Claude Code, Cursor or any coding agent — nothing to install:
@skills jeremylongshore/clay-observability