Frequently asked questions
Common questions about OpenSRE, the open-source (Apache 2.0) AI SRE agent framework for automated incident investigation and root cause analysis.
Last updated: July 13, 2026
What is OpenSRE?
OpenSRE is an open-source (Apache 2.0) AI SRE agent framework that automates incident investigation and root cause analysis. When an alert fires, it autonomously gathers context from your observability stack, tests root-cause hypotheses in parallel, and delivers a report with recommended fixes to Slack, Discord, or your incident tooling.
QuickstartIs OpenSRE really open source?
Yes. OpenSRE is licensed under Apache 2.0 and developed in the open on GitHub. The framework is written in Python, and you can read, audit, fork, and extend every part of the investigation pipeline.
View the source on GitHubHow does an AI SRE agent investigate an incident?
OpenSRE runs a structured investigation pipeline: it ingests the alert, assembles context (service ownership, recent deploys, config changes, related signals), frames the problem into testable hypotheses, then runs an agentic loop that plans queries against your logs, metrics, and diffs, executes them, and synthesizes the evidence. It stops when additional queries are unlikely to change the outcome and delivers a root-cause report with supporting evidence and recommended next actions.
How an investigation worksWhat integrations does OpenSRE support?
OpenSRE connects to 60+ tools across observability (Datadog, Grafana, Prometheus, Splunk, CloudWatch, Sentry, Honeycomb), infrastructure (AWS, Kubernetes, Helm, Argo CD), databases (PostgreSQL, MySQL, MongoDB, Redis, ClickHouse, Snowflake), incident management (PagerDuty, OpsGenie, incident.io), code hosts (GitHub, GitLab, Bitbucket), and messaging (Slack, Discord, Telegram, WhatsApp). Custom systems can be connected via MCP, APIs, and webhooks.
Integrations overviewCan I self-host OpenSRE?
Yes. OpenSRE runs locally, on-prem, or in the cloud. You install the CLI in your own environment, connect integrations from there, and your telemetry stays on your infrastructure. An enterprise setup is also available for teams that want a managed control plane.
Installation guideWhich LLM providers does OpenSRE work with?
OpenSRE is LLM-agnostic: you bring your own provider. It supports multiple LLM APIs and CLIs, configured via environment variables, and you can switch providers without changing your investigation setup.
Supported LLM providersIs there benchmark data for OpenSRE?
Yes. OpenSRE ships a benchmark harness for CloudOpsBench, a 452-scenario incident-investigation corpus, so you can run OpenSRE with your chosen LLM against the full corpus and compare results to published LLM-alone baselines.
CloudOpsBench benchmark