🌙 Post a nightly data-quality digest
Stack: Trigger: schedule (interval) · Models: your choice · Tools: warehouse/query API · Mode: single agent + memory · Output: Slack digest via channel or webhook
Who it’s for: data teams who find out about broken pipelines from angry stakeholders. What it does: every night it checks freshness, row-count deltas, and null-rate spikes across your key tables, groups the problems, and posts a short digest to Slack, or says “all clear”, before anyone opens a dashboard.
Build it
- Connect the warehouse. Connect API to a query endpoint over your warehouse (or a metrics API) for the checks you care about.
- Create the agent with: run the freshness and anomaly checks, group failures by table, and write a short digest; if nothing’s wrong, say so.
- Turn on memory so it can say “third night in a row for
orders_fct” instead of reporting each night in isolation. - Put it on a schedule: on the agent’s Schedule panel, enable it and set the interval (nightly, or hourly for tighter loops). almyty triggers it with no external cron.
- Deliver the result: publish on Slack for an in-channel post, or register an outbound webhook so almyty POSTs the digest to your endpoint when the run finishes.
- Review runs: every night’s run is in the audit trail with exactly what it queried.
Make it yours
- Swap the checks for uptime probes, SLA-breach alerts, or a KPI summary.
- Page someone via a PagerDuty tool when a check is critical, not just informational.
- Feed thresholds from a config table so analysts tune it without touching the agent.
Uses: Data · Scheduled · Slack · Memory