Calypso Multimodal RAG

Name: Calypso Multimodal RAG
Author: calypso-so

@calypso-so

Calypso Multimodal RAG について

Launch a hosted Gemini File Search true multimodal RAG layer that answers from your PDFs, docs, screenshots, charts, diagrams, help center, FAQs, and images with citations people can verify.

#rag #gemini-file-search #vector

基本情報

カテゴリ

メモリとナレッジ

トランスポート

stdio

公開者

calypso-so

投稿者

Calypso RAG

設定

以下の設定を使って、このサーバーを MCP 対応クライアントに追加してください。

{
  "mcpServers": {
    "calypso-mcp": {
      "command": "npx",
      "args": [
        "-y",
        "calypso-mcp"
      ],
      "env": {
        "CALYPSO_API_KEY": "YOUR_CALYPSO_API_KEY",
        "CALYPSO_API_BASE_URL": "https://api.calypso.so/v1"
      }
    }
  }
}

ツール

ツールは検出されませんでした

ツールは README から自動的に抽出されます。メンテナーは ## Tools という見出しの下に記載することで、このタブに反映できます。

概要

What is Calypso Multimodal RAG?

Calypso Multimodal RAG is an MCP server that brings multimodal Retrieval-Augmented Generation (RAG) to AI tools and workflows, providing a grounded answer layer for websites, AI agents, support copilots, internal knowledge assistants, and product experiences. It retrieves from actual source materials—PDFs, docs, screenshots, diagrams, charts, help content, and structured knowledge—to ensure answers remain source-backed, explainable, and production-ready.

How to use Calypso Multimodal RAG?

After configuring the Calypso MCP server in your MCP client (e.g., Claude Desktop), you can invoke three tools: calypso-rag-agent to query the RAG agent for grounded answers, calypso-upload-agent-file to upload a file into the agent store, and calypso-upload-knowledge-file to upload a file into the durable knowledge store.

Key features of Calypso Multimodal RAG

Grounded answers with retrieval-backed responses, not generic model output.
Multimodal RAG support across PDFs, docs, screenshots, diagrams, and images.
Built for AI agents and real product surfaces such as website chat and support.
Two separate upload stores: agent store for retrieval-backed RAG and knowledge store for durable indexing.
RAG-first file semantics via rag_policy on agent file attachments.

Use cases of Calypso Multimodal RAG

AI support assistants that answer from company documentation and knowledge bases.
Sales enablement copilots with grounded product information.
Internal knowledge agents retrieving from internal files and diagrams.
Website answer widgets offering searchable, citation-backed responses.

FAQ from Calypso Multimodal RAG

What is a grounded answer layer?

A grounded answer layer retrieves from real source material—such as PDFs, docs, or screenshots—rather than relying on generic model memory, so responses are source-backed, explainable, and production-ready.

What types of source material does Calypso support?

Calypso supports PDFs, documentation, help center content, screenshots, diagrams, charts, images, internal files, and structured knowledge.

How does Calypso handle file uploads?

The server provides two upload tools: calypso-upload-agent-file puts content into the agent store for retrieval-backed RAG chat, and calypso-upload-knowledge-file ingests content into a durable knowledge store for shared corpus indexing.

What makes Calypso different from other MCP RAG servers?

Calypso is designed as a multimodal grounded answer layer with reuse across multiple surfaces, retrieval over real source material, and strong fit for operational and customer-facing AI systems.

「メモリとナレッジ」の他のコンテンツ

MCP Apple Notes

RafalWilinski

Talk with your notes in Claude. RAG over your Apple Notes using Model Context Protocol.

401

Semantic Scholar MCP Server

YUZongmin

A FastMCP server implementation for the Semantic Scholar API, providing comprehensive access to academic paper data, author information, and citation networks.

158