Fandom Vision MCP

Fandom Vision is a specialized Model Context Protocol (MCP) server designed to extract, clean, and format character visual data from any Fandom/MediaWiki site.

If you've ever tried to get an LLM to generate accurate concept art or describe a character's exact outfit, you know they often hallucinate or blend different eras of a character together. Fandom Vision solves this by pulling the ground-truth visual data directly from the wiki.

How It Works

The server uses a Dual-Vector Extraction Architecture:

1. Infobox Parsing: Extracts structured physical attributes (hair color, eye color, height, species, clothing) directly from the HTML portable infoboxes.
2. Narrative Wikitext: Pulls the raw wikitext from the "Appearance" section and runs it through a robust, dynamically configured regex engine to strip out all MediaWiki artifacts, templates, and citation tags.

The result is a perfectly clean, AI-ready string of physical traits.

Available Tools

• get_raw_character_appearance: Retrieves the raw, cleaned physical appearance data for a specific character (e.g., pulling "Naruto Uzumaki" from the "naruto" domain). Perfect for maintaining visual continuity when writing or referencing characters.
• generate_vision_prompt: Takes the cleaned appearance data and automatically formats it into a highly optimized prompt string ready for Image Generation models (like Midjourney or DALL-E). Ideal for game development concept art, fan projects, or visual reference gathering.

Features

• Zero-Hallucination: Relies strictly on the wiki's current text.
• Cross-Fandom: Works across any standard Fandom wiki domain (Star Wars, Marvel, DC, Anime wikis, etc.).
• Polite Scraping: Built-in rate limiting ensures it respects Fandom's server load.