Ragstack
Give your agent search, chat, upload, and web scrape against your own serverless RAGStack knowledge base on AWS.
Overview
RAGStack MCP is an MCP server for the Build phase that connects agents to search, chat, upload, and scrape operations on a serverless AWS RAGStack knowledge base.
What is this MCP server?
- Search and chat over a managed RAGStack knowledge base
- Upload documents and scrape sources into the corpus
- Serverless AWS-backed RAGStack (HatmanStack Lambda project)
- PyPI stdio package ragstack-mcp 0.1.4 via uvx
- GraphQL API keyed from dashboard Settings
- Package ragstack-mcp version 0.1.4 on PyPI
- stdio transport via uvx
- Repository: HatmanStack/RAGStack-Lambda on GitHub
What problem does it solve?
Agents hallucinate when your docs live outside the model unless you expose a reliable RAG API through MCP.
Who is it for?
Builders who already run RAGStack on AWS and want Claude or Cursor to use the same GraphQL KB in daily coding and support flows.
Skip if: Anyone without a RAGStack tenant who wants a fully hosted turnkey RAG with zero AWS setup.
What do I get? / Deliverables
After configuration, your agent can ground replies and ingest content through RAGStack without custom MCP code.
- MCP tools for KB search and conversational query
- Document upload and scrape ingestion into RAGStack
- Agent-ready bridge to HatmanStack RAGStack-Lambda backend
Recommended MCP Servers
Journey fit
RAG wiring happens while you build the product’s knowledge and support surfaces, before you optimize distribution in launch. This is agent-facing retrieval infrastructure—tooling that lets models use your curated docs and scraped content.
How it compares
AWS RAGStack bridge over MCP, not a local-only embeddings script or generic web search skill.
Common Questions / FAQ
Who is RAGStack MCP for?
It is for developers operating a RAGStack knowledge base who want MCP-native search, chat, upload, and scrape for AI assistants.
When should I use RAGStack MCP?
Use it while building agent features or doc-heavy products that need grounded retrieval before you ship to customers.
How do I add RAGStack MCP to my agent?
Set RAGSTACK_GRAPHQL_ENDPOINT and RAGSTACK_API_KEY from Dashboard → Settings, then register uvx ragstack-mcp (0.1.4) as a stdio MCP server in your client.