Claude Visual Communications Plugin

Claude-Visual-Communications-Plugin is a Claude Code plugin for planning and prompt-engineering AI-generated visuals, including images, diagrams, and video, for whitepapers, blog posts, technical documents, and other long-form content. It was originally built as a workspace template named Claude-Visual-Communications-Space and later repackaged as an installable plugin. Project state is stored in the plugin data-storage path rather than in the host repository, with each project organised into source, visuals, and prompts directories alongside a project.yaml file and an ideation document.

The plugin provides six skills: onboard-visual-project for setting up a project's name, type, platform, audience, and style; ideate-visuals for proposing five to eight visual concepts from source material; generate-visual-prompt for converting a concept into model-specific prompts for Flux, SDXL, or video models, with variations and negative prompts; list-visual-projects for reviewing project status; run-fal-generation for executing saved prompts through the fal-ai MCP server; and visual-style-reference for platform resolution guidelines and style-modifier vocabulary. Only run-fal-generation requires the optional fal-ai MCP server, and the remaining skills operate without it. The tool naming can be remapped through a local configuration file for users who access fal through a different MCP gateway.

The plugin is suited to technical writers, developers, and content teams who produce long-form documentation and want a structured workflow for creating accompanying illustrations or diagrams with generative AI. It is released under the MIT License by Daniel Rosehill.