Alert Decision Layer (CLI)

A small, opinionated alert triage engine you can run as a CLI or scheduled job.

It ingests alerts (e.g., exported from Prometheus Alertmanager or PagerDuty as JSON), applies an architecture-driven decision model, and outputs:

• A decision for each alert: page, ticket, suppress, or aggregate.
• A short reason explaining why the decision was made.
• A consolidated Markdown report and JSON report you can paste into your incident tooling.

The goal is to move from alert fatigue to a simple, explicit "alert decision layer" that can later be extended with AI explanations.

Features

• Normalizes alert payloads into a simple schema.
• Applies rules based on:
  • severity (critical, warning, info)
  • environment (prod, staging, etc.)
  • service risk profile (from a small services.yml file)
  • alert history hints (flapping / high volume per fingerprint)
• Emits human-readable Markdown and machine-readable JSON.

Quickstart

git clone https://github.com/AutoShiftOps/alertdecider.git
cd alertdecider
python -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt

# Example
python -m alertdecider-agent --alerts examples/alerts.json   --services examples/services.yml   --history  examples/history.json   --out-dir out

cat out/decision_report.md