Ragora

What is Ragora?

Ragora is a hosted cloud MCP server that lets you search your knowledge bases from any AI assistant using the Model Context Protocol. It requires no local installation—just configure your AI tool with the server URL and an API key.

How to use Ragora?

Sign up at ragora.app and create an API key at Settings > API Keys. Then connect your AI tool (Claude Desktop, Claude Code, Cursor, VS Code, Windsurf, or any MCP client) by adding the server URL https://mcp.ragora.app/mcp and your API key as an Authorization: Bearer header.

Key features of Ragora

• Hosted cloud service, no local installation required.
• Dynamic tools generated per knowledge base collection.
• Hybrid search with dense (E5-Large) and sparse (BM25).
• Streamable HTTP transport for MCP.
• Multi-tenant with API key scoping.
• Tools: discover_collections, search, search_collection.

Use cases of Ragora

• Search across all your knowledge bases with a single natural language query.
• Look up specific policies or documentation in a named collection.
• Discover available workspaces and their contents with discover_collections.
• Retrieve information about a specific topic from a collection.
• List available documentation versions for a collection.

FAQ from Ragora

What runtime or dependencies do I need?

None. Ragora is a hosted cloud service—you only need an internet connection and an API key.

Where is my data stored?

Your data is stored in Ragora's cloud infrastructure, using Postgres for metadata and Qdrant for vector embeddings.

What transport and authentication does Ragora use?

Ragora uses Streamable HTTP transport (HTTP POST to the MCP endpoint) and authenticates via a Bearer token in the Authorization header.

What should I do if I get a "Tool not found" error?

Restart your AI app after configuration changes and verify your API key starts with sk_live_.

What should I do if I get an "Unauthorized" error?

Check that the Authorization: Bearer header is formatted correctly and that your API key hasn't expired.