ContextFS is a Model Context Protocol (MCP) server that empowers AI assistants with direct, file-first access to your local document collections, enabling advanced search and knowledge retrieval.

How does ContextFS integrate with AI assistants?

ContextFS acts as an MCP server, allowing AI models to leverage its tools for browsing collections, performing full-text searches with boolean operators, and reading documents directly from your local filesystem.

What document types does ContextFS support?

ContextFS offers extensive format support for PDFs (via poppler-utils), Markdown, plain text, and other common formats through its extensible and configurable filter system.

Does ContextFS require a database or RAG infrastructure?

No, ContextFS is a 'no database, no RAG' solution. It leverages `ugrep` for direct, powerful full-text search across your filesystem, simplifying deployment and reducing overhead.

What security measures does ContextFS implement?

ContextFS ensures security with read-only access to documents, path validation, command sandboxing, and whitelist enforcement for shell filters, protecting your local data.

ContextFS

Name: ContextFS
Author: RomanShnurov

byRomanShnurov

0•

Developer Tools

Learning & Documentation

Data Science & ML

Provides AI assistants with direct, file-first access to local document collections, offering full-text search, fuzzy matching, and hierarchical context for various document types.

ContextFS is a Model Context Protocol (MCP) server designed to empower AI assistants with intelligent access to local document collections. It eliminates the need for complex database infrastructure by leveraging powerful command-line utilities like ugrep for lightning-fast full-text search, including boolean operators and fuzzy matching, and poppler-utils for robust PDF text extraction. Documents are organized into hierarchical collections mirroring your filesystem, allowing for global, collection-specific, or document-level search scopes. With a strong emphasis on security, ContextFS operates in a read-only environment, preventing document modification while offering features like path validation and command sandboxing, making it an ideal secure bridge between LLMs and your file-based knowledge base or codebase.

Key Features

01Hierarchical collection organization with global, collection, and document-level search scopes

02Support for PDF, Markdown, plain text, and other formats via an extensible system

03Read-only access, path validation, and command sandboxing for enhanced security

040 GitHub stars

05Full-text search with boolean operators, fuzzy matching, and context highlighting

06Direct filesystem integration with no database or RAG infrastructure required

Use Cases

01Providing AI with context-aware search capabilities across large local PDF libraries or code repositories

02Integrating local documentation, research papers, or codebases directly with AI assistants

03Building a secure, file-first knowledge base that LLMs can query for information retrieval