O

Onyx

Open Source AI Platform - AI Chat with advanced features that works with every LLM

32.3k
Stars
+90
Stars/month
1506
Commits (90d)
10
Releases (6m)

Star Growth

+270 (0.8%)estimated from velocity
31.4k32.2k32.9kJul 4Oct 2

Overview

Onyx is a knowledge/context layer that connects to 50+ applications to index and surface team knowledge for AI agents. It provides features like agentic RAG, custom agents, web search, external actions via MCP, a secure sandbox for code execution, and voice mode. It supports all major LLM providers and emphasizes self-hosted deployments for data sovereignty.

Deep Analysis

Key Differentiator

Provides a unified, low-latency internal knowledge representation for agents, replacing iterative MCP searches with instant context fetching.

⚡ Capabilities

  • • Agentic RAG
  • • Custom AI agent creation
  • • Multi-step deep research
  • • Web search integration
  • • Secure code sandbox
  • • External actions via MCP
  • • Voice mode interaction
  • • Artifact generation

🔗 Integrations

50+ applications for knowledge indexingSerper, Google PSE, Brave, SearXNG for web searchFirecrawl/ExaOllama, LiteLLM, vLLMAnthropic, OpenAI, Gemini

✓ Best For

  • ✓ Teams wanting an AI coworker with internal knowledge
  • ✓ Building AI agents with custom knowledge subsets
  • ✓ Complex research workflows requiring ground truth
  • ✓ Self-hosted deployments requiring data sovereignty

✗ Not Ideal For

  • ✗ End-users seeking a simple generic chatbot
  • ✗ Teams needing only public LLM knowledge without internal context
  • ✗ Non-technical users requiring fully managed SaaS

⚠ Known Limitations

  • ⚠ Requires deployment and connection to knowledge sources
  • ⚠ Complexity of managing self-hosted infrastructure
  • ⚠ Agent coordination may be less suitable for trivial queries

Alternatives

See all 8 Onyx alternatives →

Compare Onyx

Maintain Onyx?

Show your live rank in your README, or put Onyx in front of every visitor to AgentoolRank.