Home / Engine / Use cases

A knowledge base MCP server for your documents

Connect StackGrep to Claude, Claude Code, Cursor or Codex once, and your agent can search every collection you've loaded: exact text, regex, metadata filters, exact counts, and the full document when it needs it. Nothing to host.

Join the waitlist See the docs

Why the usual way falls short

Local MCP servers only see files on one laptop
Building your own search server means indexing, hosting and auth
Most retrieval tools return snippets an agent can't verify

How it looks

# Claude Code
claude mcp add --transport http stackgrep https://stackgrep.com/mcp

# tools your agent gets
list_collections · search_collection · count_collection · get_document

What your agent asks, and what it calls

“Which of our runbooks mention the payments database?”search_collection(query: "payments-db")

“How many support macros still say “7 days”?”count_collection(query: "7 days")

“Open the macro so I can edit it”get_document(id: "macros/refund-window")

Questions

How do agents sign in?

MCP clients that support OAuth (Claude, Claude Code, Cursor and others) open a StackGrep sign-in and consent page; each connected client gets its own revocable credential. Others can send an API key header.

Can an agent change my data?

The collection tools are read-only: list, search, count, read. Writes go through the REST API with an API key.

Does it work with the web index too?

Yes. The same server also has tools to search the source of the websites we've indexed and to fetch a live page as Markdown.

More use cases

Hosted grep for your AI agents Exact counts for RAG and agents Exact match search for AI agents RAG without embeddings Grep an S3 bucket: search the contents of every file A regex search API that doesn't scan everything Find every contract that says exactly this

Give your agents exact search

StackGrep Engine is onboarding teams from the waitlist. Bring your documents or your bucket.

Join the waitlist Read the docs