Reading
Harpyx is a multi-tenant, self-hostable Retrieval-Augmented Generation (RAG) platform designed specifically for enterprise private document libraries. It supports virus scanning, multi-provider LLM integration, project-level conversation isolation, as well as complete audit logs and security controls.
Section 01
Harpyx is a multi-tenant, self-hostable Retrieval-Augmented Generation (RAG) platform designed specifically for enterprise private document libraries. It supports virus scanning, multi-provider LLM integration, project-level conversation isolation, as well as complete audit logs and security controls.
Section 02
Current RAG (Retrieval-Augmented Generation) products on the market generally have three core issues: First, they are locked into specific large model providers, lacking flexibility; Second, they require uploading documents to third-party cloud services, making data privacy difficult to guarantee; Third, most are single-tenant toy projects that cannot adapt to the organizational structure and permission management needs of real enterprises. Harpyx was born to solve these problems.
Section 03
Harpyx is a multi-tenant, self-hostable document intelligence platform built on .NET 10, using a dual-service architecture (Web frontend + Worker backend) and deployed via Docker Compose. It allows organizations to upload private documents, index them into searchable vector embeddings, and conduct conversational interactions via user-selected large language models.
Section 04
Harpyx's document ingestion process is very comprehensive: uploaded files first undergo virus scanning, then are stored in object storage and queued for asynchronous parsing. The system supports decompression of multiple container formats (ZIP, RAR, 7z, tar.gz, MSG, EML), as well as extraction, chunking, and embedding of PDFs, Office documents, RTF, EPUB, HTML, plain text, images (OCR recognition), and structured files (CSV, JSON, XML, YAML).
Section 05
The platform supports user-configurable API keys and is compatible with OpenAI, Anthropic Claude, and Google Gemini. All keys are encrypted using AES-256-GCM when stored at rest. Chat, embedding, and OCR models can be independently overridden at the workspace or project level, providing great flexibility.
Section 06
Harpyx uses a four-layer organizational architecture: platform-level tenants group users into workspaces; workspaces contain multiple projects; each project includes documents, prompts, and chat sessions based on its own documents. This design ensures strict data isolation while supporting cross-project collaboration.
Section 07
Platform-level roles include Admin, Operator, Reviewer, and ReadOnly; tenant members have an independent role model. User access is controlled via whitelist, and all operations are auditable.
Section 08
Harpyx provides instance-level quota management, which can limit the usage of tenants, workspaces, projects, documents, storage, APIs, OCR, and RAG. Resource management can be achieved without a commercial tier.
Keep going with more reads from the same topic.
Nornir MCP Server is an enterprise-level server based on the Model Context Protocol (MCP). It seamlessly integrates large language models (such as Claude) with the Nornir network automation framework, supporting natural language orchestration for multi-vendor network devices (Cisco, Arista, Juniper, etc.), and providing production-grade features like a dual-engine architecture (NAPALM + Netmiko), intelligent filtering, and a secure sandbox.
Bibliothèque Française LLM is a structured indexing and annotation project for French public domain literature designed specifically for large language models (LLMs). It integrates multiple authoritative sources such as DraCor, Common Corpus, and Wikisource, providing metadata indexing categorized by genre, author, and era, as well as in-depth annotations for dramatic texts (including characters, lines, stage directions, etc.). Its aim is to enable LLMs to efficiently read and understand classic French literary works.
Splinter is a minimalist, high-performance key-value (KV) and vector storage system enabling zero-latency inter-process communication via shared memory and atomic operations. With only 766 lines of core code, it supports millions of operations per second and 768-dimensional vector storage, offering a new architectural approach for local LLM inference and data-intensive applications.
A complete cloud-native AI application development guide for beginners, building a simple generative AI chatbot using Amazon EC2, Apache, Python CGI, and Amazon Bedrock, covering architecture design, IAM permission configuration, security best practices, and cost optimization suggestions.