Practical overview
When it fits
Use KaaS when you want to transform files and transcripts into an editable Markdown knowledge base and query it through an LLM, while operating the service yourself.
Who it is for
- Individuals organizing scattered notes and reference documents into a searchable personal wiki.
- Teams compiling meeting transcripts and role-specific knowledge into reusable documentation.
- Developers connecting a self-hosted knowledge base to MCP-compatible AI clients.
Documented uses
- Compile raw material into categorized Markdown articles and indexes.
- Ask questions against the compiled wiki and trace answers to named source articles.
- Make a knowledge base available to MCP clients through an authenticated endpoint.
Requirements & limits
The project documents CLI and Docker deployment. The CLI installer lists Linux amd64/arm64 and macOS arm64; Docker images are listed for Linux amd64 and arm64. An OpenAI-compatible LLM API key is required for the documented setup paths. The repository specifies the MIT license.
- KaaS requires configuring an OpenAI-compatible LLM endpoint and API key for the documented setup.
- Its output is a tree of Markdown articles, which can be browsed and edited as documents.
- The README describes a four-phase LLM compilation pipeline and retrieval from whole pages, rather than chunk-and-embed retrieval.
- Remote MCP is an optional service configuration and supports bearer-token authentication in the documented setup.
Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.
Start KaaS with Docker
Run the KaaS service in Docker with an OpenAI-compatible LLM endpoint configured.
Before you start
- Docker and a supported Linux amd64 or arm64 environment.
- An API key and model endpoint compatible with the OpenAI API.
Steps
Run the documented Docker command with port 8080 mapped, a local data directory mounted at /app/data, and LLM_API_KEY, LLM_BASE_URL and LLM_MODEL set.
Open http://localhost:8080 after the container starts.
Expected outcome
A running KaaS service with its web interface, configured to call the selected LLM endpoint.
Guide limitations
- The documented image targets Linux amd64 and arm64.
- The edge image tracks main; the README recommends pinning a version for deployments that depend on a specific release.
Enable remote MCP access
Expose a running KaaS instance to an MCP client using bearer-token authentication.
Before you start
- A KaaS service run with Docker or kaas serve.
- An MCP client able to connect to the service host.
Steps
Set KAAS_MCP_ENABLED=true in the service environment.
Set KAAS_MCP_TOKEN to a secret token.
Configure the MCP client to use http://:8080/mcp and send Authorization: Bearer .
Expected outcome
The MCP client can address the KaaS MCP endpoint using the configured bearer token.
Guide limitations
- The guide covers MCP connection configuration, not client-specific setup details.
- The service must be reachable at the host and port supplied to the MCP client.
Sources & review
Source record details
- Source availability
- Active
- Sources checked
- 12 October 2026
- Original website title
- bybit-exchange/kaas · Official homepage title · retained in its source language