Skill

nano-tech-diagrams

Create and edit tech diagrams using the nano-tech-diagrams MCP server (Nano Banana 2 via Fal AI). Supports text-to-image, image-to-image, whiteboard cleanup, and 28+ style presets.

npx claudepluginhub danielrosehill/claude-code-plugins --plugin image-production

Tool Access

This skill uses the workspace's default tool permissions.

Preview

This skill uses the **nano-tech-diagrams** MCP server to create and transform tech diagrams using Fal AI's Nano Banana 2 model.

SKILL.md

Similar Skills

cache-components

139.3k

Guides Next.js Cache Components and Partial Prerendering (PPR): 'use cache' directives, cacheLife(), cacheTag(), revalidateTag() for caching, invalidation, static/dynamic optimization. Auto-activates on cacheComponents: true.

cache-components

mcp-builder

124.2k

Guides building MCP servers enabling LLMs to interact with external services via tools. Covers best practices, TypeScript/Node (MCP SDK), Python (FastMCP).

9 files

anthropics-skills-13

Stats

Stars1

Forks0

Last CommitApr 27, 2026

Actions

View Source View Plugin View on GitHub View README

Help us improve

Share bugs, ideas, or general feedback.

Nano Tech Diagrams

This skill uses the nano-tech-diagrams MCP server to create and transform tech diagrams using Fal AI's Nano Banana 2 model.

MCP Server Location

Always use mcp__jungle-shared__nano-tech-diagrams__* (on ubuntuvm).

Do not use jungle-local. Its schema claims to accept local filesystem paths, but the process runs in an environment that cannot read the workstation's filesystem, so every local-path call fails with ENOENT. The shared instance plus the staging endpoint below is the supported path.

File Staging (CRITICAL)

The MCP server cannot read workstation file paths directly. For any tool that accepts an image_path parameter with a local file, stage it to the residence S3 (mcp-staging bucket) and pass the presigned URL as image_path — the MCP accepts http(s) URLs directly.

Staging workflow:

Upload the local file and get a presigned GET URL:
```
URL=$(s3-stage /local/path/to/image.jpg)
```
Pass $URL as image_path to the MCP tool.
Objects auto-expire after 1 day (bucket lifecycle); presigned URL TTL defaults to 1 hour (--expires to override).

Batch staging:

For multiple files, call s3-stage for each (can be parallelised), collect the URLs, then call the MCP tools.

When staging is NOT needed:

text_to_image — no input image required
When image_path is already an http:// or https:// URL (Fal media, any public host)

Saving output to the workstation

The MCP returns a fal.media URL. Do not rely on the download_to parameter — it writes to the ubuntuvm process's filesystem, not the workstation. Download the returned URL yourself:

curl -sSL -o "/absolute/workstation/path/output.png" "<fal.media URL>"

Available Tools

Tool	Purpose
`list_styles`	Show all 28+ visual style presets (key, name, category, default aspect ratio)
`list_diagram_types`	Show all diagram type presets (network, flowchart, mind map, etc.)
`whiteboard_cleanup`	Clean up a whiteboard photo into a polished diagram
`image_to_image`	Transform an existing image into a styled tech diagram
`text_to_image`	Generate a diagram from a text description (no input image)

Defaults

Model: Nano Banana 2 (baked in, not configurable)
Resolution: 1K (default)
Reasoning: Minimal (default)
Output format: PNG
Aspect ratio: auto (pass as parameter to override)

Workflow

For whiteboard cleanup:

Stage the image with s3-stage <path> to get a presigned URL
Optionally ask about style preference — default is clean_polished
Call whiteboard_cleanup with the presigned URL as image_path
If domain-specific terms are present, pass them as dictionary_words for accurate spelling
Download the returned fal.media URL to the workstation (see "Saving output" above)

For image-to-image transformation:

Stage the image with s3-stage <path> to get a presigned URL
Determine what transformation is needed: style change, diagram type conversion, or custom prompt
Call image_to_image with at least one of: prompt, style, or diagram_type, and the presigned URL as image_path
Common combos: style=blueprint + diagram_type=network_diagram
Download the returned fal.media URL to the workstation

For text-to-image generation:

Get the user's description of what diagram they want
Pick an appropriate diagram_type if it matches a preset
Pick a style if the user has a visual preference
Call text_to_image — no staging needed (no input image)
Download the returned fal.media URL to the workstation

Style Categories

Professional: clean_polished, corporate_clean, hand_drawn_polished, minimalist_mono, ultra_sleek, blog_hero
Creative: colorful_infographic, comic_book, isometric_3d, neon_sign, pastel_kawaii, pixel_art, stained_glass, sticky_notes, watercolor
Technical: blueprint, dark_mode, flat_material, github_readme, photographic, terminal_hacker, visionary
Retro & Fun: chalkboard, psychedelic, mad_genius, retro_80s, woodcut
Language: bilingual_hebrew, translated_hebrew

Diagram Types

Infrastructure: network_diagram, cloud_architecture, kubernetes_cluster, server_rack
Software: system_architecture, microservices, api_architecture, database_schema
Process: flowchart, decision_tree, sequence_diagram, state_machine, pipeline
Conceptual: mind_map, wireframe, gantt_chart, comparison_table, org_chart

Notes

For batch processing, call tools sequentially to avoid API rate limits
The dictionary_words parameter helps with domain-specific terminology (e.g., ["Kubernetes", "Proxmox", "PostgreSQL"])
Aspect ratio options: auto, 1:1, 4:3, 3:4, 16:9, 9:16, 3:2, 2:3, 21:9, 9:21
Resolution options: 0.5K, 1K (default), 2K, 4K
The mcp-staging S3 bucket has a 1-day object lifecycle; presigned URLs default to 1-hour TTL