KaaS
A self-hosted AI knowledge base that builds an editable Markdown wiki, connects to Ollama or cloud APIs, and gives agents cited answers through MCP.
KaaS is a self-hosted AI knowledge base for personal research and teams preserving knowledge across role changes. It turns notes, documents and transcripts into wiki articles you can read, edit and query.
- Wiki compilation: An LLM organizes incoming material into articles and merges overlapping information, producing readable pages rather than a store of embedded text fragments.
- Cited answers: Ask questions in the web interface and receive answers that name the relevant wiki articles; retrieval uses whole pages as context.
- Text, files and URLs: Add pasted text, uploaded documents or web pages, with incremental compilation that avoids reprocessing the existing collection for each addition.
- Editable Markdown: Keep articles as plain files on disk, edit them by hand, and track or review changes with Git.
- Agent access: Claude Code, Codex and other MCP-compatible agents can query the wiki and receive cited answers through local or remote connections.
The application and wiki run on your own machine or server. KaaS connects to OpenAI-compatible backends, including Ollama, vLLM, OpenAI, DeepSeek and Azure OpenAI. Model requests go to the backend you choose, so a cloud connection sends the content used for those requests outside your machine.
KaaS is open source under the MIT license. Docker builds support Linux on AMD64 and ARM64, while CLI builds support Linux and Apple Silicon macOS. The Docker image includes the web interface, backend and AI engine.
