Quarry FAQs

Question 1

What is Quarry and what is its primary function?

Accepted Answer

Quarry is a powerful tool designed to ingest diverse document types, including scanned PDFs, images, and text files. It processes them using OCR, creates local vector embeddings, and enables semantic search, making your internal knowledge accessible to Large Language Models (LLMs) via the Model Context Protocol (MCP) or a command-line interface (CLI).

Question 2

What document types can Quarry process, including those requiring OCR?

Accepted Answer

Quarry can process a wide range of documents: scanned PDFs, mixed-format PDFs, various image formats (PNG, JPG, TIFF, BMP, WebP), and text files (TXT, Markdown, LaTeX, DOCX), as well as raw text. It utilizes AWS Textract for automatic page classification and Optical Character Recognition (OCR) on image-based and scanned content.

Question 3

How does Quarry facilitate semantic search locally?

Accepted Answer

Quarry performs local vector embeddings using the `snowflake-arctic-embed-m-v1.5` model. These embeddings are stored efficiently in LanceDB, a fast, local vector database, which allows for powerful semantic search capabilities without needing an external database.

Question 4

How does Quarry integrate with LLMs and what is the Model Context Protocol (MCP)?

Accepted Answer

Quarry seamlessly integrates with LLMs like Claude Code and Claude Desktop by exposing a comprehensive MCP server. The Model Context Protocol (MCP) allows LLMs to interact with Quarry's specialized tools for semantic search, document ingestion, management, and status checks, enabling them to query and understand your document repository.

Question 5

Does Quarry require cloud services or an external database for its operation?

Accepted Answer

Quarry primarily operates with local vector storage via LanceDB, eliminating the need for an external database. However, it leverages AWS Textract for its OCR capabilities, which requires AWS credentials and an S3 bucket for temporary file handling during the OCR process.

Quarry

Quarry

主要功能

使用案例

主要功能

使用案例