Self-hosted Postfix log search correlated by queue ID, plus a delivery-health dashboard built from your own mail logs.
One tool for the two ways mail-ops actually works: the fast lookup when something breaks, and the quiet watch on whether delivery is healthy.
Someone reports a delivery problem. Arrive with an address, domain, or subject, get the matched log entries fast, leave. Lines are parsed and stitched by queue ID into structured records with per-recipient status, shown raw and formatted.
A glanceable read on aggregate delivery health over time. A background scanner rolls log data into hourly, daily, and weekly tiers in a libSQL store, so trends stay fast across months of mail.
No agents, no log shipping, no external store. It reads the mail.log the box already writes and turns it into search and signal.
The signal that matters, readable in seconds. The same six KPIs and trend charts you get at /stats, rolled up hourly to weekly so months of mail stay fast.
Scrape /metrics with Prometheus (durable counters survive restarts and aggregate across workers), drop in the bundled Grafana dashboards over the Infinity datasource, and point uptime checks at /health. Pull the same numbers over HTTP from the stats CLI or a scoped read-only key, and let an embedded MCP server expose search and stats to your agents.
Search is for when you already know something broke. Everything else, the tool watches for you, then puts a readable summary where you will actually see it.
Set a threshold or let it learn the baseline. Scope to the domains that matter. Route to email or a webhook.
A branded HTML delivery-health digest in your inbox, on your schedule.
An LLM-written summary from current and prior stats, plus one-click export of the live dashboard.
Twelve prebuilt themes, chosen for mail-ops eyes and long shifts. The accent, temperature, and contrast change. Delivered green, bounced red, deferred amber stay fixed in every one.
No source checkout. Pull the compose file and config template, set one secret, and start the app container with its sqld sidecar. It runs on the box that already collects the mail.
/, dashboard at /stats:v3.1.1 for reproducible deploys$ mkdir postfix-insights && cd postfix-insights# a working dir$ curl -fsSL https://raw.githubusercontent.com/Xodus-CO/postfix-insights/main/docker-compose.prod.yml -o docker-compose.yml# compose file$ curl -fsSL https://raw.githubusercontent.com/Xodus-CO/postfix-insights/main/.env.example -o .env# config template$ sed -i "s/^CSRF_SECRET=$/CSRF_SECRET=$(openssl rand -hex 32)/" .env# set the secret$ docker compose up -d# go→ http://<host>:8080