Pheme — Observability Intelligence Layer

A non-invasive intelligence layer that plugs into existing Prometheus, Thanos, and Grafana stacks to deliver anomaly detection, service topology inference, and AI-driven incident analysis—without migration, re-instrumentation, or vendor lock-in.

Features

• Anomaly Detection: Holt-Winters ETS trained on existing Thanos history
• Service Topology: Inferred from kube-state-metrics
• AI Incident Analysis: Pluggable LLMs (Ollama, Anthropic, OpenAI, Bedrock)
• Grafana Native: Service Map, Incident Timeline, Anomaly Alerts, AI Chat plugins
• Read-Only: Additive layer; zero impact on existing stack

Quick Start

Prerequisites: Go 1.21+, Python 3.11+, Docker, Docker Compose

git clone https://github.com/phemehq/pheme.git
cd pheme

docker-compose -f deploy/docker-compose.yml up -d
go test ./... && pytest python/tests/
go run ./cmd/bridge &

Visit http://localhost:3000 (Grafana) to see Pheme plugins.

Architecture

Existing Stack (read-only)
  ↓
Ingestion → Intelligence Core → AI Layer → Grafana Plugins

See docs/ARCHITECTURE.md for details.

Contributing

See CONTRIBUTING.md for setup, code style, and workflow.

Security

For security vulnerability reporting, see SECURITY.md.

License

Apache License 2.0 — See LICENSE.

Status: Early development (pre-MVP)