Nano-Tech-Diagrams Plugin
April 30, 2026 · View on GitHub
Generate, transform, and clean up tech diagrams and whiteboard photos via Fal AI's Nano Banana 2 model. Wraps the nano-tech-diagrams MCP server with seven skills and a curated prompt library across 5 diagram families and 28+ visual styles.
Skills
generate-tech-diagram— text-to-image diagram generation (network, flowchart, K8s, mind map, etc.)restyle-diagram— transform an existing image into a different visual stylecleanup-whiteboard— turn whiteboard / sketch photos into polished diagramslist-diagram-styles— browse the 28+ visual style presetslist-diagram-types— browse the diagram-type presetsdiagram-prompt-library— curated, known-good prompt templates by family and stylewhiteboard-to-publishable— full pipeline: photo → cleaned diagram → comparison + caption
Requirements
This plugin is a wrapper. The actual generation runs through the nano-tech-diagrams MCP server, which must be installed and configured with a Fal AI API key.
1. Install the MCP
claude mcp add nano-tech-diagrams -- npx -y nano-tech-diagrams-mcp
Or pin via npm globally and reference the binary directly. See nano-tech-diagrams-mcp on GitHub.
2. Set the Fal AI API key
Export FAL_KEY (or FAL_AI_API_KEY) in the environment Claude Code launches the MCP from.
3. (Optional) Install the image-annotation plugin
The whiteboard-to-publishable orchestrator can optionally use it for side-by-side comparison images and redaction. Without it, the orchestrator degrades gracefully.
Installation
# User scope (always available):
claude plugins install nano-tech-diagrams@danielrosehill
# Or project scope (run inside the target repo):
claude plugins install nano-tech-diagrams@danielrosehill --scope project
Data storage
Generated images and projects default to:
${CLAUDE_USER_DATA:-${XDG_DATA_HOME:-$HOME/.local/share}/claude-plugins}/nano-tech-diagrams/
You can always pass an explicit output path to any generation skill.
Credits
- Generation engine: Nano-Tech-Diagrams-MCP
- Prompt library substrate: Tech-Diagram-Image-Gen-Prompts (vendored snapshot)
- Sample corpus: Sample-Whiteboards
- Underlying model: Fal AI's Nano Banana 2
License
MIT