#multimodal

Valossa Assistant is a multimodal video-AI MCP server built on a decade of video understanding. Unlike video MCPs that stop at search or analysis, Valossa goes end to end: ingest a video, understand it (summary, chapters, named entities, keywords, visual scenes, IAB categories, sensitive-content flags), ask it questions conversationally, get transcripts and captions, surface the best soundbites and highlights — and export a finished MP4 clip with burned-in subtitles, smart-crop reframing (e.g. 9:16) and caption styling. Add the server, sign in for a free trial, and your agent has a complete video production pipeline. EU-built, credit-based, self-serve.

MCP server for Calypso’s multimodal RAG layer. Supports agent-store uploads, knowledge-store ingestion, and retrieval-backed responses workflows with file-aware rag_policy semantics.

A structured-document editor for AI agents. SuperDocs gives your AI 21 MCP tools and 4 workflow prompts to make section-precise edits — bold a specific paragraph, replace a single table cell, restructure a heading — without disturbing surrounding content. Tables, borders, alternating row shading, fonts, and inline styling all survive AI edits AND round-trip exports across .docx, PDF, HTML, Markdown, and RTF. Other capabilities: pre-signed URL upload/download (no context bloat for files >100KB), compact response mode for editing 100-page documents efficiently (~140× token reduction), multimodal vision on attachments, human-in-the-loop approval for sensitive edits, and multi-language editing across 16+ languages. Free plan: 500 ops/month, no credit card required.

Multimodal Cultural Intelligence Infrastructure for AI Agents. The only MCP server that analyzes text, images, and video for cultural risks across 200+ markets. 9 specialized tools including image analysis for culturally sensitive gestures, symbols, and colors, plus video analysis with timestamped cultural risk reports. Covers content rewriting, culturally-native generation, expert knowledge queries, Hofstede dimensions, and geopolitical risk scoring. Free tier available.

Turn any language model into a multimodal powerhouse that can generate images, music, videos, 3D models and more on the fly. Rostro's tools are designed to be used by language models from the ground up, expanding capabilities with minimal context bloat.

Process and prepare your multimodal medical data with natural language!

A iOS/MacOS Swift MCP Client using voice interacting with python MCP servers both natively

Digimon Engine — Multi-Agent, Multi-Player Framework for AI-Native Games and Agentic Metaverse

A multimodal mcp server