cyberneticlibrary

Generate marketing images with AI

image-promptingskillsetup L2465
blockrunai/blockrun-mcp
What it does

Generate images with legible text and precise layouts

Best for

Creating marketing posters, UI mockups, product shots, or any image where on-image typography must render exactly as specified without AI-generated slop.

Inputs
  • · [object Object]
  • · [object Object]
  • · [object Object]
Outputs
  • · [object Object]
Requires
  • · blockrun_image MCP client
  • · OpenAI GPT Image 2 API (best for text)
  • · Black Forest Flux 1.1 Pro (photoreal)
  • · Google Nano Banana (fast/cheap)
Preconditions
  • · blockrun_image tool available in MCP
  • · USDC wallet funded (~$0.01–0.06 per image)
  • · Prompt structured in 5-section format
Failure modes
  • · Vague prompt → model generates unexpected variation
  • · Text prompt ambiguous → model renders wrong words
  • · Composite from multiple refs → model conflates sources
  • · Shallow DOF accidentally applied to full frame
  • · Extra text or watermark appears despite 'no extra text' constraint
Trust signals
  • · 5-section framework proven to prevent prompt failure
  • · Anti-slop rules documented (visual facts over praise adjectives)
  • · Text wrapping rules explicit (quotes, CAPS, letter-by-letter spelling)
  • · GPT Image 2 empirically best for legible text vs. other models
  • · Iterative edit workflow documented (preserve constraints every turn)