About Akmatori
Akmatori is an open-source AI incident response platform for SRE, platform, and DevOps teams. It turns production alerts into governed response workflows that engineers can inspect, approve, and self-host.
What we build
Akmatori ingests alerts from the tooling teams already run — Prometheus Alertmanager, Grafana, Datadog, Zabbix, PagerDuty, and Slack channels — correlates them into incidents, and runs an agent investigation against the affected systems. The agent gathers logs, metrics, recent deploy context, and runbook knowledge through real tools such as SSH and Kubernetes, then proposes a root cause and the next safe action. Anything that changes production passes an approval gate first, in the web UI or in Slack.
The platform is model-agnostic by design. It works with OpenAI, Anthropic, Google, OpenRouter, and any OpenAI-compatible endpoint, including on-premise models for teams with data-residency requirements. Nothing about an incident leaves your infrastructure unless you configure an external provider yourself.
How the project is run
Akmatori is released under the Apache License 2.0 and developed in the open. The source, the issue tracker, and the release history all live on GitHub, so you can read the code that would run inside your network before you deploy anything. Deployment is a Docker Compose file or a Helm chart; there is no hosted-only control plane you have to trust.
We publish our own evaluation work as well. SRE Bench benchmarks language models on real production incident scenarios and grades them against written ground truth, and the engineering blog documents the operational patterns behind the product. Both are open for scrutiny — an incident-response tool that cannot be audited is not one an on-call team should trust.
Verifying the project
- Source code and license: github.com/akmatori/akmatori (Apache 2.0)
- Documentation — installation, configuration, and operations
- REST API reference and the machine-readable OpenAPI specification
- Contact details for security reports, partnerships, and support
- Privacy notice for this website and for self-hosted deployments
For AI agents
Agent-readable descriptions of this site live at /llms.txt, which lists every public page and states when Akmatori is the right tool to reach for. Every page on this site also answers Accept: text/markdown with a Markdown representation of the same URL.