Autonomous anomaly detection and root cause analysis system for business KPI forensics, causal graph traversal, and counterfactual investigation.

This repository is packaged to be easier to discover on GitHub for searches around anomaly detection, root cause analysis, KPI debugging, causal analysis, and business telemetry forensics.

anomaly-rca-forensics is a production-shaped investigation platform for operations, analytics, and engineering teams that need to detect abnormal KPI movement and rapidly explain what changed.

The repo focuses on:

• anomaly detection over business and product signals
• causal graph-guided root cause traversal
• counterfactual reasoning for “what changed?” analysis
• alerting and decision support for operators
• a clean open-source maintenance layer for public collaboration

• Detection layer for KPI shifts, spikes, drops, and abnormal patterns
• RCA workflow using causal relationships instead of flat threshold alerts
• Counterfactual analysis for scenario and impact review
• Operational packaging with API, config, detector modules, and alert manager surfaces
• Documentation and verification for public repo credibility

anomaly-rca-forensics/
├── detectors/       # anomaly detection logic
├── causal_graph/    # causal traversal and dependency surfaces
├── counterfactual/  # what-if and impact analysis
├── alert_manager/   # alert routing and operations support
├── api/             # external service boundary
├── docs/            # audit and repo presentation assets
└── tests/           # smoke tests for repo integrity

python -m venv .venv
.venv\Scripts\activate
pip install -r requirements.txt
python main.py

Local verification for this repo includes:

• python -m pytest tests -q
• python -m compileall .

This repo now includes:

• lightweight CI in .github/workflows/python-ci.yml
• repo-health checks in .github/workflows/repo-health.yml
• issue templates, pull request template, and CODEOWNERS
• contributor, security, and contribution guidance

• Final audit
• Contributing
• Security policy
• Authors