What is knowledge-engine?

Knowledge Engine is a self hosted tool for people who use an AI assistant like Claude every day for writing, research, or building things, and want a persistent, searchable record of that work. It combines two ideas into one system: a human readable wiki made of plain markdown files, and a machine friendly search layer, so the same information can be browsed by a person or searched instantly by an AI. Every document, conversation, or work session you feed into it gets turned into a markdown wiki page with metadata, links to related pages, and automatically extracted entities such as people, companies, or products mentioned in it. Those pages can be browsed in the built in web dashboard or in an external tool like Obsidian since they are just plain files in a folder. Search queries check both the curated wiki pages and the raw underlying documents, then merge the results with links back to their sources. The dashboard includes several other tabs beyond search and the wiki browser: a health tab that checks whether the wiki and the underlying search index have drifted out of sync, a tool inventory tab that lists every AI agent, script, or tool you have built or installed, and a monthly summary tab that tallies hours, documents produced, and tools built over the month. A command line report script can also generate a summary for a custom date range, such as the last week or a specific client's work, and print it as markdown. The project is written mostly in Python with an HTML dashboard, installs with pip, and includes demo data you can ingest to try it out before using it on real material. An optional companion library called Memvid can be added for faster search across very large document sets, but the wiki layer works on its own for smaller collections. A separate sister repository handles automated backups of the whole system.