SRE Whisperer

SRE Whisperer is a full-stack demo system for autonomous IT operations. A FastAPI backend simulates faults, performs causal inference with networkx, emits live SSE updates, and drives an SRE agent. A Next.js 14 frontend renders the war-room dashboard, live metric charts, a causal dependency graph, agent reasoning, and the human approval flow.

Repository Layout

• backend/ FastAPI app, causal engine, agent loop, mock services, and Chroma helpers
• frontend/ Next.js 14 App Router dashboard and hidden control panel
• scripts/ seeding and scenario validation scripts
• PLANS.md milestone plan derived from the blueprint

Quick Start

Backend

1. Create a virtual environment and install the backend requirements:

   python -m venv .venv
   .\.venv\Scripts\Activate.ps1
   pip install -r backend/requirements.txt

2. Optional LLM setup for the LangChain path:

   ollama pull mistral:instruct

3. Start the backend:

   uvicorn backend.main:app --reload --port 8000

Frontend

1. Install dependencies:

   Set-Location frontend
   npm.cmd install

2. Start the dashboard:

   npm.cmd run dev

3. Open http://localhost:3000.

Validation

Backend tests

python -m pytest backend/tests -q
python scripts/test_scenarios.py

Frontend tests

Set-Location frontend
npm.cmd run test

Notes

• sentence-transformers is intentionally configured with device='cpu' to avoid RTX 4050 VRAM pressure and match the blueprint guidance.
• If langchain_ollama or Ollama is unavailable, the agent falls back to a deterministic local reasoning path so the demo remains runnable.