Onyx

Onyx is the knowledge and context layer for your team and AI agents. It connects to your applications to index and surface knowledge from 50+ applications, and it protects data sovereignty through flexible self-hosted deployments.

Beyond search, Onyx offers custom agents with their own subsets of knowledge and instructions, a multi-step deep research flow, web search, external actions through MCP, a code sandbox and downloadable artifacts. It supports all major LLM providers, both self-hosted ones such as Ollama, LiteLLM and vLLM and proprietary ones such as Anthropic, OpenAI and Gemini.

Onyx can be reached from a web and desktop app, a Slack and Discord bot, an MCP server for clients such as Claude Code, a Chrome extension and an embeddable widget. The self-hosted deployment is air-gappable, with the document index, database and processing inside your own services. It runs on Docker, Kubernetes, and Helm or Terraform. Standard Onyx includes indexing; Onyx Lite is a lightweight chat UI that uses under 1GB of memory and cannot index documents. Onyx Community Edition is available freely under the MIT license, and a one-command installer is provided.