ragstack — Claude Code MCP server: setup
Connects an AI assistant to an AWS knowledge base for document search, uploads, OCR, transcription, and chat with retrieved context.
claude mcp add --env RAGSTACK_GRAPHQL_ENDPOINT=<RAGSTACK_GRAPHQL_ENDPOINT> --env RAGSTACK_API_KEY=<RAGSTACK_API_KEY> --transport stdio ragstack -- uvx ragstack-mcpWhat ragstack does
ragstack exposes an AWS-hosted knowledge base through MCP. An assistant can search stored material, accept new files, and use content processed with OCR or speech transcription. Supported inputs include PDFs, images, Office files, HTML, CSV, JSON, XML, EML, EPUB, video, and audio.
The MCP package runs as uvx ragstack-mcp and uses a GraphQL endpoint plus an API key. The underlying stack uses AWS Lambda, S3, DynamoDB, and Bedrock. Claude Code uses claude mcp add; Cursor and GitHub Copilot use their MCP configuration format.
The server cannot connect without RAGSTACK_GRAPHQL_ENDPOINT and RAGSTACK_API_KEY. The API key is available from Dashboard → Settings.
What it helps with
- search a private knowledge base
- add files for later retrieval
- extract text from PDFs and images
- transcribe recorded video and audio
- retrieve answers with source references
Example prompts for the agent
- Search my knowledge base for authentication documentation.
- Find the API setup details in the uploaded PDFs.
- Locate the video and audio sections that discuss authorization.
How to install ragstack
Platform CLI
claude mcp add --env RAGSTACK_GRAPHQL_ENDPOINT=<RAGSTACK_GRAPHQL_ENDPOINT> --env RAGSTACK_API_KEY=<RAGSTACK_API_KEY> --transport stdio ragstack -- uvx ragstack-mcpReplace <…> with your own values. uvx runs the command named after the package; if the README names another one, use uvx --from ragstack-mcp <command>
Platform CLI
codex mcp add ragstack --env RAGSTACK_GRAPHQL_ENDPOINT=<RAGSTACK_GRAPHQL_ENDPOINT> --env RAGSTACK_API_KEY=<RAGSTACK_API_KEY> -- uvx ragstack-mcpReplace <…> with your own values. uvx runs the command named after the package; if the README names another one, use uvx --from ragstack-mcp <command>
Config file
{
"mcpServers": {
"ragstack": {
"command": "uvx",
"args": [
"ragstack-mcp"
],
"env": {
"RAGSTACK_GRAPHQL_ENDPOINT": "<RAGSTACK_GRAPHQL_ENDPOINT>",
"RAGSTACK_API_KEY": "<RAGSTACK_API_KEY>"
}
}
}
}Replace <…> with your own values. uvx runs the command named after the package; if the README names another one, use uvx --from ragstack-mcp <command>
Config file
{
"servers": {
"ragstack": {
"type": "stdio",
"command": "uvx",
"args": [
"ragstack-mcp"
],
"env": {
"RAGSTACK_GRAPHQL_ENDPOINT": "<RAGSTACK_GRAPHQL_ENDPOINT>",
"RAGSTACK_API_KEY": "<RAGSTACK_API_KEY>"
}
}
}
}Replace <…> with your own values. uvx runs the command named after the package; if the README names another one, use uvx --from ragstack-mcp <command>
Platform CLI
gemini mcp add -e RAGSTACK_GRAPHQL_ENDPOINT=<RAGSTACK_GRAPHQL_ENDPOINT> -e RAGSTACK_API_KEY=<RAGSTACK_API_KEY> ragstack uvx ragstack-mcpReplace <…> with your own values. uvx runs the command named after the package; if the README names another one, use uvx --from ragstack-mcp <command>
Step by step, with commands for each platform: How to add an MCP server to Claude Code, Codex, Cursor, VS Code, Gemini CLI and more
Questions about ragstack
How do I install ragstack in Claude Code?
Run claude mcp add ragstack -- uvx ragstack-mcp, then configure access with `claude mcp add --env RAGSTACK_GRAPHQL_ENDPOINT=<RAGSTACK_GRAPHQL_ENDPOINT> --env RAGSTACK_API_KEY=<RAGSTACK_API_KEY> --transport stdio ragstack -- uvx ragstack-mcp`.
What does ragstack do?
It connects Claude Code to an AWS knowledge base through MCP, allowing the assistant to search documents, add material, and answer questions from stored content.
What does ragstack need to connect?
Set RAGSTACK_GRAPHQL_ENDPOINT and RAGSTACK_API_KEY. The API key comes from Dashboard → Settings.
