Local Knowledge Base Manager

Problem

A local knowledge base usually breaks into several disconnected steps: ingestion, chunking, vectorization, retrieval, tagging, and export. Depending on a cloud embedding API also adds network dependency and privacy concerns. This skill targets users who need to maintain offline document knowledge, supporting PDF, TXT, MD, DOCX, JSON, CSV, and HTML, while using local models for semantic search.

How it works

The core flow is import → local embedding → semantic search → tag filtering → stats/export. It uses Ollama's nomic-embed-text model instead of a cloud API, so after setup, knowledge base operations stay local. Day-to-day use can be triggered with natural-language actions for adding documents, searching, managing tags, and viewing statistics. It also supports Markdown and JSON import/export, which helps with backup, migration, or integration into existing workflows. Platform support covers macOS, Linux, WSL2, and Windows via WSL2, depending on local setup and manual configuration.

Limitations

It is better suited as a personal or small-team local document search tool focused on offline retrieval and knowledge organization, not a general-purpose document management system. Multi-user permissions, audit logs, real-time sync, complex policy controls, and enterprise-grade backup require additional components. Users should confirm that Ollama and the embedding model can run locally, and remember that vector search quality depends on document chunking and the embedding model.