CHERENKOV Architecture¶

This document describes the architectural principles and design patterns of the CHERENKOV project.

1. The Trident of Truth¶

CHERENKOV operates on the "Trident of Truth" framework: - MEISSNER: Network enforcement and proxy layer. Controls external access. - ABLATION: Data redaction and anonymization. Protects sensitive information. - TOKAMAK: Sandboxed validation. Ensures untrusted code and scanners are executed safely.

2. Clean Architecture¶

The project strictly enforces a Clean Architecture pattern, isolating concerns across layers: - Persistence Layer (ResultStore): Manages all state and data storage. Must not be accessed directly by CLI or external scripts. - API Layer (orchestration_api.py): The single entry point for routing, business logic, and orchestration. All external clients (CLI, Web Dashboards) MUST use this layer. - CLI/Client Layer: Thin wrappers that interact only with the API layer.

3. Agent Governor and Cognitive Routing¶

CHERENKOV leverages an Agent Governor pattern to route tasks among various internal AI agents dynamically. Agents are assigned specific roles and rules, and their outputs are moderated by the governor to ensure safety, alignment, and execution of the given flow.

4. Scanners Plugin Architecture¶

Scanners must inherit from the BaseScanner class (src/cherenkov/core/base_scanner.py). This guarantees: - Auto-discovery via the ScannerRegistry. - Standardized execution through the scan() interface. - Consistent output formatting (ScanResult).

graph TD
    A[Component] --> B[Subcomponent]
    B --> C[Implementation Detail]