Try Aurora Live • Quick Start • Documentation • Integrations • Website • Changelog
What's New
- Artifacts — Persistent agent-maintained documents in Monitor, continuously updated as investigations progress
- Actions — Automated post-RCA workflows (generate postmortems, open fix PRs, notify Slack) triggered on investigation completion
- AWS Bedrock Support — Use Claude, Titan, and other Bedrock models via IAM auth
- Fly.io Connector — Investigate incidents on Fly.io infrastructure
- CloudBees Enterprise — Operations Center + Feature Management connector
- Kubeconfig Upload — Connect on-prem Kubernetes clusters without a cloud provider
- CloudWatch Alarm Webhooks — Ingest AWS CloudWatch alarms directly as incidents
- Extensibility Hooks — Gate LLM calls, enforce seat limits, and customize behavior with lifecycle hooks
See the full CHANGELOG for all releases.
Why Aurora?
When an alert fires at 3 AM, your on-call engineer spends 30-60 minutes doing the same thing every time: checking dashboards, running kubectl commands, reading logs, correlating deployments, and searching Slack history.
Aurora does all of that autonomously. It receives the alert, spins up AI agents that investigate across your entire stack, and delivers a structured RCA by the time you open your laptop.
| Without Aurora | With Aurora |
|---|
| Engineer paged, context-switches | Alert auto-triaged in background |
| 30-60 min manual investigation | AI agents investigate in parallel |
| Knowledge siloed in individuals | Investigation reasoning captured |
| Postmortem written days later | Postmortem auto-generated |
| Same failure, different engineer | Knowledge base grows over time |
Click to watch the full demo
How well does it work?
We test it in the open. Aurora Lab breaks a Kubernetes cluster in 19 known ways, sends Aurora only the alert, and grades the investigation against the injected root cause. Every run is published, including the ones Aurora gets wrong.
- Benchmark: Aurora and HolmesGPT on the same 19 faults, with the losses. K8sGPT takes no alert and scans the namespace, so its runs are reported separately, not scored against the other two
- Runbooks: one per lab fault (OOMKilled, ImagePullBackOff, Service with no endpoints, and more), built from the real runs
Features
Agentic Investigation
AI agents dynamically select from 30+ tools. They run kubectl, aws, az, and gcloud in sandboxed Kubernetes pods, query logs, check deployments, and correlate data — all autonomously.
AI Code Fixes
Aurora doesn't just find root cause — it suggests fixes and can generate pull requests with the remediation.
Incident Dashboard
Ingest alerts from PagerDuty, Datadog, Grafana, New Relic, OpsGenie, incident.io and more. Every alert auto-triggers a background investigation.
Auto-Generated Postmortems
Detailed reports with timeline, root cause, impact assessment, and remediation steps. Export directly to Confluence, Notion, or SharePoint.
Infrastructure Knowledge Graph
Visualize your entire infrastructure as a dependency graph. When an incident occurs, Aurora traces blast radius across services and providers.
More capabilities: Knowledge Base RAG • Multi-Cloud (AWS, Azure, GCP, OVH, Scaleway, Cloudflare) • Any LLM (OpenAI, Anthropic, Gemini, Vertex AI, OpenRouter, Ollama) • Terraform/IaC Analysis • MCP Server (Cursor, Claude Desktop, Windsurf) • Org-level Command Policies • SigmaHQ Guardrails • NeMo Input Rail
Quick Start
Get Aurora running locally in under 5 minutes:
git clone https://github.com/arvo-ai/aurora.git && cd aurora
make init # Generate secure secrets
nano .env # Add your LLM API key (OpenRouter, OpenAI, etc.)
make prod-prebuilt # Pull prebuilt images and start
Open http://localhost:3000. The first user to register becomes admin.
[!TIP]
Aurora works without any cloud provider accounts. The LLM API key is the only external requirement. Connectors are optional.
# Get the auto-generated root token
docker logs vault-init 2>&1 | grep "Root Token:"
# Add it to .env
echo "VAULT_TOKEN=hvs.your-token-here" >> .env
# Restart to connect services to Vault
make down && make prod-prebuilt
make prod-prebuilt VERSION=v1.2.3
make prod-local
Deploy
| Method | Best for |
|---|
make prod-prebuilt | Local evaluation, single-node |
| Helm chart | Production Kubernetes (GKE, EKS, AKS) |
| Air-tight bundle | Air-gapped / restricted networks |
Kubernetes (Helm)
helm repo add aurora https://raw.githubusercontent.com/Arvo-AI/aurora/gh-pages
helm repo update
helm show values aurora/aurora-oss > my-values.yaml
# Edit my-values.yaml, then:
helm install aurora-oss aurora/aurora-oss -n aurora --create-namespace -f my-values.yaml
Also available via OCI: oci://ghcr.io/arvo-ai/charts/aurora-oss
This is the beginning of the README. Read the rest at the original. Read the full README ↗
