Karzoun Sentinel AI

What it is

Offline-first LLM evaluation & regression testing for AI apps and agents. Prompt injection, groundedness, secret redaction, JSONL suites and CI quality gates.

Open-source evaluation and regression testing for LLM applications and AI agents. SentinelAI is a Python-first toolkit for evaluating AI outputs before they reach production. It turns prompts, responses, grounding context, and future agent traces into repeatable test cases that can run locally or inside CI.

Engineering characteristics

• Automated tests
• Containerised
• Continuous integration
• Dependency manifest
• Documentation set
• Open licence
• Release pipeline
• Security policy

Testing

The repository contains an automated test suite that runs as part of its checked-in workflow.

Security

The repository publishes a security policy describing how to report vulnerabilities.

Deployment

The repository defines its own build and deployment automation.

Documentation

The repository ships a dedicated documentation set beyond the README.

Topics