Skip to content

Constitutional AIOpsAutonomous infrastructure management with a safety layer

A dual-agent system that annotates telemetry, finds root cause, and proposes remediation. Every action passes a constitutional safety gate and, when uncertain, waits for human approval. Bring your own OpenAI-compatible LLM endpoint.

Constitutional AIOps architecture. Telemetry feeds a fast agent and a reasoning agent backed by graph-episodic memory, then every proposed action passes the constitutional safety gate, which routes it to an audited auto-run, human approval, or an alert.

What it is ​

Constitutional AIOps ingests logs, metrics, and traces, uses a fast agent to annotate and classify them, and a reasoning agent to correlate incidents, explain root cause, and draft remediation. Nothing is executed blindly. A constitutional validator scores each proposed action, and the authorization matrix decides whether it runs automatically, waits for a human, or only raises an alert.

The recommended way to run it is the lite profile: the backend and frontend plus your own OpenAI-compatible LLM endpoint. No GPU, no bundled models, no Neo4j required.

bash
git clone https://github.com/Partha-dev01/Constitutional-AIOps.git constitutional-aiops
cd constitutional-aiops
cp .env.example .env          # set your LLM endpoint (both agents may share one)
docker compose -f docker/docker-compose.lite.yml --env-file .env up -d
# Frontend  http://localhost:3000
# Backend   http://localhost:8000/docs

Continue with the Overview, or jump to Self-Hosting for every deployment option.

License ​

Constitutional AIOps is released under the GNU Affero General Public License v3.0. If you run a modified version as a network service, you must offer its users the corresponding source of your modified version.

Released under the AGPL-3.0 License.