Barghest

OpsAI is an AI assistant for log analysis and guided incident remediation. Ops teams investigate incidents across logs scattered over many systems. Today that means manual correlation, remediation that depends on one person's know-how, and logs pasted into public AI tools without any control over sensitive data. OpsAI puts the whole loop in one tool. It ingests logs from 10 connectors (file, journald, Docker, Kafka, Loki, Elasticsearch, syslog, Kubernetes, Windows Event Log, Azure Monitor) and filters them with its own query language (OQL). It mines message templates with Drain3, masks sensitive data with regex and Presidio, and only after a payload preview sends the anonymized evidence to a Meta Muse model. The model returns a schema-validated diagnosis: hypotheses with confidence and evidence references, plus a step-by-step remediation plan with risk class, verification and rollback for each command. Approved steps run in an execution engine (dry-run, four-eyes approval, time windows, local or SSH, rollback). Every action is recorded in a hash-chained audit log. Desktop UI, web UI and REST API run in one process (Python 3.12, NiceGUI). Ships as .deb, AppImage, a Docker image, a Helm chart and ArgoCD/Kustomize GitOps.