Seed Loom

Seed Loom is a lightweight repository for turning creative signals into structured systems: briefs, design-language extractions, essay engines, art-product pipelines, implementation specs, validation gates, and agent handoff instructions.

This repo currently hosts the Signal-to-System Pipeline as an executable workflow scaffold.

Core idea

Most AI-assisted creative work fails because it stops at output. Seed Loom treats every input as a seed that should become a traceable system:

signal -> interpretation -> system -> artifact -> product -> ecosystem

What is included

• docs/SIGNAL_TO_SYSTEM_PIPELINE.md — the operating manual
• agent/AGENT_MODE_HANDOFF.md — compact instructions for ChatGPT, Agent Mode, Cursor, Copilot, Replit, Kimi, or Perplexity
• schemas/project-intake.schema.json — structured project intake schema
• templates/ENGINE_RUN_CARD.md — one-page execution card for any project
• figma/FIGMA_BOARD_SPEC.md — Figma/FigJam mapping plan
• .github/workflows/signal-validation.yml — basic repository validation workflow
• scripts/check_pipeline.py — validates required files and JSON schema syntax

The four engines

1. Visual Grammar Engine — extracts motifs, materials, palette, symbolic language, composition logic, prompt grammar, and negative constraints.
2. Mythic Essay Engine — turns notes and research into publishable longform essays with voice rules, section architecture, cover prompts, tags, and social excerpts.
3. Art-to-Product Engine — turns generative art collections into metadata, captions, alt text, SKU logic, Gumroad copy, print/PDF layout, and launch assets.
4. Spec-to-Build Engine — translates aesthetic intent into buildable implementation specs for React, Astro, WebGL, GLSL, R3F, Framer, and interactive prototypes.

Out-of-the-box implementation idea

This repo is designed to become a living evidence ledger, not a static doc dump. Every project can get a run card with:

• source signal
• selected engine chain
• generated artifacts
• validation evidence
• failure modes
• handoff target
• product/publishing state

That means the work becomes cumulative. Each artifact becomes searchable, reusable, testable, and shippable.

Recommended workflow

1. Open a new issue using the project intake format.
2. Choose the engine chain.
3. Generate a run card.
4. Produce outputs into /docs, /templates, /products, or /case-studies.
5. Validate using GitHub Actions.
6. Mirror the visual map in Figma/FigJam.

Status

v0.1 scaffold — ready for expansion into docs, product pack, Figma map, and working agent templates.