platform-operations/platform-knowledge

Platform Knowledge

Content verification, fact-checking, and knowledge management for the Lilith platform.

Structure

platform-knowledge/
├── crystal-ai/     Python AI — Crystal auditor, training pipeline, CLI
├── shared/         TypeScript config — service ports, URLs, shared constants
└── README.md

Components

crystal-ai/

The Crystal knowledge AI system. Scans platform documentation, generates training data, and provides the Crystal CLI for interactive fact verification.

• Python packages: lilith-platform-knowledge (library) + crystal-cli (CLI)
• Infrastructure: docker-compose.yml (crystal-redis + crystal-postgres)
• Orchestration: run script (start/stop/status/infra/logs)
• Models: Fine-tuned GGUF models for knowledge verification
• Systemd: Production service units for API and training

shared/

TypeScript configuration shared across the monorepo. Service ports, URLs, and constants for the knowledge-verification API.

Related

• Knowledge Platform (upstream): ~/Code/@applications/@ml/knowledge-platform/
• Client package: @lilith/knowledge-verification-client (published to Verdaccio)
• Admin package: @lilith/knowledge-verification-admin
• Crystal TUI launcher: tooling/run/cli/commands/crystal.ts (./run crystal)