Skip to content

InfraSigns

Stop reacting to alerts. Start knowing what's coming.

InfraSigns connects to your Prometheus and delivers a daily AI-generated digest — overall health, anomalies, and recommended actions — straight to Telegram. When incidents fire, it correlates the alert storm into a single root-cause summary instead of forwarding 50 individual notifications.

How it works

Every morning at the configured time, InfraSigns:

  1. Scrapes your Prometheus for active alerts and key metrics
  2. Sends the data to your LLM (OpenAI or Anthropic)
  3. Delivers a concise summary to Telegram

Example output: 

Infrastructure Health Digest — production

• Overall status: HEALTHY
• CPU at 18.3% — nominal
• Memory at 61.2% — healthy
• Disk at 45.8% — healthy
• No active alerts

No action required. System is performing normally.

When critical alerts fire, InfraSigns correlates them and sends one message:

Incident Summary — production

Root cause: NodeDown on prod-db-1 (firing since 14:23 UTC)
Impact: Database writes failing — API returning 503

Immediate actions:
  1. Check prod-db-1: ssh prod-db-1
  2. Review recent deployments: git log --oneline -5
  3. Failover to replica if node is unrecoverable

Key features

  • Zero alert fatigue — one message per incident, not one per firing rule
  • Root cause detection — correlates alerts by instance and job
  • Daily digest — proactive health summary every morning
  • Self-hosted — your data never leaves your infrastructure
  • Single binary — no dependencies, runs anywhere Go runs
  • Docker-ready — docker compose up and you're done

Try it now

git clone https://github.com/infrasigns/infrasigns
cd infrasigns
go run ./cmd/demo

No API keys or Prometheus required. See the Getting Started guide for production setup.