Context-Harbor: Self-hosted knowledge layer for AI teams and agents

Hi Product Hunt 👋 I built ContextHarbor after repeatedly seeing the same knowledge infrastructure rebuilt for every AI project: document ingestion, semantic chunking, indexing, retrieval, access control, and integration with AI agents. ContextHarbor brings these capabilities together in a self-hosted platform. Teams can deploy it with Docker, keep documents and vectors on infrastructure they control, organize knowledge into access-controlled projects, and use it through a Web UI, REST API, CLI, or remote MCP. The search pipeline combines semantic retrieval with keyword matching, which helps it understand meaning while still finding exact terms such as error codes, class names, product terminology, and requirement identifiers. I’m currently looking for engineering teams to pilot ContextHarbor in real-world environments. I’d especially value feedback on deployment, retrieval quality, MCP workflows, and the types of internal knowledge your teams need to make accessible to AI applications. I’d love to hear what you’re building and what you currently use for your knowledge layer.