Lugh

Lugh is a 7-stage adaptive learning pipeline — a separate Guildhall workflow with its own vault at /content/Lugh/.

This note serves as a bridge between the Guildhall vault and the Lugh vault. For full Lugh documentation, see the Lugh vault directly.

What it is

A local LLM-powered system that takes a topic string (“Understanding Design Patterns”) and produces an adaptive podcast-style course with Feynman-style tutoring and gated advancement. Named after Lugh Lámhfhada, the Samildánach — “skilled in all arts.”

The pipeline

  1. Topic seeding and curriculum scoping (Stage 0)
  2. Pre-assessment Feynman diagnostic
  3. Per-episode research and synthesis via RAG
  4. Script generation
  5. TTS rendering
  6. Listening phase
  7. Feynman tutor assessment → gates advancement to next episode

The key architectural decision: Episode 2 only generates after passing Episode 1’s assessment.

Relationship to Guildhall

Lugh is a workflow configuration within Guildhall. It uses the same infrastructure:

  • Model registry for Ollama model assignments per stage
  • Job runner for pipeline execution on Blackthorn
  • Audit trail for tracking learner progress and model outputs
  • Memory layer for accumulating teaching patterns across courses

Relationship to Quorum

Quorum is Lugh’s design review board, not part of its runtime:

  • Stage 0 curriculum review
  • Gate calibration (pass/fail thresholds)
  • Course Zero design (“Understanding Critical Thinking”)

Lugh vault location

/Users/christopherhagmann/Projects/Personal/knowledge/content/Lugh/