image-gen

L1 Improvement

• Recast the image generation SOP using Prompt Architect clarity and Skill Forge structure-first guardrails.
• Added provider selection rules, confidence ceilings, and memory tagging.
• Clarified integration with visual-art-composition and safety constraints.

---

LIBRARY-FIRST PROTOCOL (MANDATORY)

Before writing ANY code, you MUST check:

Step 1: Library Catalog

• Location: .claude/library/catalog.json
• If match >70%: REUSE or ADAPT

Step 2: Patterns Guide

• Location: .claude/docs/inventories/LIBRARY-PATTERNS-GUIDE.md
• If pattern exists: FOLLOW documented approach

Step 3: Existing Projects

• Location: D:\Projects\*
• If found: EXTRACT and adapt

Decision Matrix

Match	Action
Library >90%	REUSE directly
Library 70-90%	ADAPT minimally
Pattern exists	FOLLOW pattern
In project	EXTRACT
No match	BUILD (add to library after)

---

STANDARD OPERATING PROCEDURE

Purpose

Select the best image provider (local or API) for the request, generate assets with structured prompts, and deliver reproducible outputs with safety and cost visibility.

Trigger Conditions

• Positive: requests for illustrations, banners, UI mockups, diagrams, or concept art.
• Negative: vector/icon design requiring manual tools; route to design specialists.

Guardrails

• Structure-first docs maintained; scripts/resources referenced.
• Provider selection uses evidence: availability, cost, latency, privacy.
• Respect safety filters and license terms; avoid unsafe content.
• Confidence ceilings declared for prompt-to-image fit; cite provider used.
• Memory tagging for prompts/results to enable reuse.

Execution Phases

1. Intent & Constraints – Capture use case, size/aspect, style, privacy/cost limits. Pull structured prompt from visual-art-composition if available.
2. Provider Selection – Choose SDXL (local, private), DALL·E (quality), Replicate (cost), or custom; log rationale.
3. Setup/Validation – Ensure provider credentials/models ready; note latency/VRAM needs.
4. Generation – Run prompt with seed/size/options; generate 1–N variants.
5. Review & Safety – Check for content/safety violations and quality fit; rerun if needed.
6. Delivery – Save paths, provider, settings, and confidence line; store in memory.

Output Format

• Provider + rationale, prompt used, parameters (size, seed, steps), and file paths.
• Quality/safety notes and follow-ups.
• Confidence: X.XX (ceiling: TYPE Y.YY) with source of certainty (observation vs inference).

Validation Checklist

• Constraints captured; provider rationale logged.
• Credentials/models verified; no unsafe content.
• Outputs saved with deterministic params when possible.
• Memory tagged; confidence ceiling declared.

Integration

• Pipeline: combine with visual-art-composition for structured prompts.
• Scripts: scripts/multi-model/image-gen/cli.py with --list, --setup <provider>, and generation commands.
• Memory MCP: skills/tooling/image-gen/{project}/{timestamp} for prompts/results.

Confidence: 0.70 (ceiling: inference 0.70) – SOP aligned to Prompt Architect clarity and Skill Forge delivery.