AIOps Cockpit Uses an AI Agent to Diagnose Incidents with Human Approval

Summary

AIOps Cockpit is an open-source, self-hosted SRE control room that uses an AI agent to investigate infrastructure incidents. Users describe a problem, such as checkout API errors or a restarting database, and the agent uses real tools to inspect logs and metrics before proposing a fix. The project identifies Claude 3.7 as the language model used for these investigations. It does not let the AI apply changes directly: instead, the agent generates a unified diff that a human must review and approve, and the requester cannot approve their own request under a required two-person rule. Additional safeguards include append-only audit logs, prompt-injection defenses, a strict tool allowlist, and destructive actions disabled by default. The interface includes a 3D spatial topology map that represents infrastructure health and failure states. It also provides live CPU, RAM, disk, and Docker container telemetry through Socket.IO. Optional MediaPipe hand gestures can control the camera. The project requires Node.js 20 or later and pnpm, and can be started with the repository's frontend and backend together after environment setup.