Overview

Penrose-Lamarck is a learning platform designed to preserve human cognitive sovereignty in the age of AI. The architecture prioritizes:

• Clean separation of domain logic from infrastructure (Clean Architecture).

• GenAI orchestration with explicit tool boundaries via MCP servers.

• Retrieval-augmented generation for verifiable reasoning.

• Continuous evaluation, observability, and governance.

Core Principles

• Domain isolation: The learning domain is framework-agnostic.

• Ports and adapters: All I/O is abstracted by interfaces.

• Immutable state: Domain data is treated as immutable by default.

• Safety and traceability: Every model response is evaluated, traced, and attributable to a source.

Container Runtime

The platform standardizes on Podman for containerization:

• Daemonless design improves security and auditability.

• Rootless-first execution aligns with least-privilege standards.

• Strong OCI compliance and systemd integration fit enterprise operations.

Nerdctl is a strong alternative in containerd-native environments, but Podman provides more secure defaults for multi-service developer and CI workloads.