LATTICE GenAI Asset Pipeline

Manage plant asset generation from ComfyUI with quality-gated promotion into a Pixeltable registry.

3|Updated May 11, 2026
One-click install
npx skills add https://github.com/JeromyJSmith/lattice-platform --skill lattice-genai-asset-pipeline
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: LATTICE GenAI Asset Pipeline
Source: https://github.com/JeromyJSmith/lattice-platform/tree/main/.agents/skills/lattice-genai
Command: npx skills add https://github.com/JeromyJSmith/lattice-platform --skill lattice-genai-asset-pipeline

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

It prevents low-quality 2D-to-3D plant outputs from polluting your model registry by enforcing audit, tracking, evaluation, and threshold-based promotion for every generation and asset registration.

Core Features & Use Cases

  • Asset inventory + orphan detection: Ensures every tracked .glb under assets/plants/ has a corresponding registry row and flags missing-file inconsistencies.
  • ComfyUI training job dispatch + tracking: Creates and updates lattice/genai/comfyui_jobs entries through pending/running/completed/failed lifecycle states.
  • Quality-scored promotion with LOD checks: Computes quality_score, blocks promotion below 0.7, and validates LOD distribution expectations so placeholders and detail assets remain balanced.
  • Single-writer safety for training dispatch: Uses a lock (/tmp/vwbridge-genai.lock) to ensure only one training job runs at a time.

Quick Start

Run the genai cycle to audit, evaluate, and quality-gate promotion for plant assets using the repository’s autoresearch entry point.

Frequently Asked Questions about LATTICE GenAI Asset Pipeline

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I prevent low-quality 2D-to-3D plant assets from polluting a model registry?▼

Quality-gated promotion blocks any generated 3D plant asset with a quality_score below 0.7 from entering the model registry. This thresholded promotion rule enforces audit parity between filesystem assets and registry rows, preventing low-quality outputs from polluting the production environment.

How do I track ComfyUI training jobs through pending, running, and completed states?▼

Tracking ComfyUI training jobs involves dispatching them with single-writer safety using a lock file to ensure only one job runs at a time. The system then updates registry entries through pending, running, completed, and failed lifecycle states for reproducible tracking.

What is the best way to audit filesystem assets and fix orphan records in a GenAI pipeline?▼

Auditing filesystem assets and fixing orphan records requires checking every tracked .glb file under your assets directory against corresponding registry rows. This process flags missing-file inconsistencies and maintains audit parity between the filesystem and the registry.

Can I use Pixeltable for model registry management with ComfyUI generated assets?▼

Yes, you can use Pixeltable for model registry management by backing the registry with it to track .glb plant assets. This setup enforces dev-to-prod promotion rules, LOD distribution health checks, and audit parity for ComfyUI generated assets.

Why does LOD distribution health checking matter for 3D plant asset generation?▼

LOD distribution health checking matters for 3D plant asset generation because it validates that placeholders and detail assets remain balanced across LOD 100 to 300 levels. This ensures generated models meet expected quality standards before thresholded promotion to the registry.