Neura Market
An Open-source platform for building and orchestrating AI agents logo

An Open-source platform for building and orchestrating AI agents

Free
FreeFree tier
Type
Open Source
Visit Website GitHub (1 stars)

About An Open-source platform for building and orchestrating AI agents

Obsidian AI is an open-source platform for building, deploying, and orchestrating AI agents, multi-agent teams, and automated workflows from a single visual interface. It supports multiple LLM providers including OpenAI, Anthropic, Google Gemini, Ollama, OpenRouter, and any OpenAI-compatible endpoint. The platform offers a drag-and-drop canvas for creating multi-agent teams, sequential pipelines, and parallel DAG workflows without writing code. Key features include agent building with custom prompts and tool attachments, MCP protocol support, human-in-the-loop (HITL) review, knowledge bases with RAG, long-term agent memory, prompt optimization, WhatsApp integration, and production-ready security (JWT auth, TOTP 2FA, AES encryption, RBAC). It is self-hosted, ensuring data stays on the user's infrastructure, and includes a secrets vault, session history, execution traces, and an admin panel.

Key Features

Multi-Provider LLM Support: Connect to OpenAI, Anthropic, Google, Ollama, OpenRouter, or any OpenAI-compatible endpoint
Visual Agent Builder: Create agents with custom system prompts, model selection, and tool attachments via a drag-and-drop canvas
Multi-Agent Teams and Workflow Automation: Build sequential pipelines and parallel DAG workflows
Real-Time Chat Playground: Test and interact with agents in real time
MCP Protocol Support: First-class Model Context Protocol for connecting external tools and services
Human-in-the-Loop (HITL): Add human review steps to agent workflows
Knowledge Bases and RAG: Integrate retrieval-augmented generation for context-aware responses
Long-Term Agent Memory: Persist agent memory across sessions
Agent Versioning and Rollback: Version control for agents with rollback capability
Eval Harness and Regression Testing: Test agent performance and prevent regressions

Pros & Cons

Pros
  • Completely visual interface – no SDKs, boilerplate, or glue code required
  • Supports multiple LLM providers without vendor lock-in
  • Self-hosted open-source – data never leaves your servers
  • Production-ready security features (JWT, 2FA, encryption, RBAC)
  • MCP-native for easy integration with external tools and services
  • Drag-and-drop canvas for creating complex multi-agent workflows
  • Includes advanced features like eval harness, prompt optimization, and versioning
Cons
  • Requires self-hosting setup and maintenance
  • Community-driven support; no official enterprise support
  • May have a learning curve for users unfamiliar with Docker and backend configuration

Best For

Building and managing AI agents without codingOrchestrating multi-agent teams for complex tasksAutomating business workflows with AICreating self-hosted, privacy-preserving AI solutionsDeveloping AI applications with multiple LLM providersIntegrating AI agents with external tools via MCP protocolEnabling human review in AI decision-making pipelines

FAQ

What is Obsidian AI?
Obsidian AI is an open-source platform for building, managing, and orchestrating AI agents. It provides a visual interface so you don't need to write code or use SDKs to create and run agents.
Which LLM providers are supported?
It supports OpenAI, Anthropic, Google Gemini, Ollama, OpenRouter, and any OpenAI-compatible endpoint (like LM Studio or vLLM).
How does security work?
Obsidian AI includes JWT authentication, TOTP two-factor authentication, AES end-to-end encryption, a Fernet secrets vault, role-based access control, and rate limiting.
Can I self-host Obsidian AI?
Yes, it is fully self-hosted. You run it on your own infrastructure and your data never leaves your servers.
What is the Model Context Protocol (MCP)?
MCP is a protocol for connecting external tools and services to your AI agents. Obsidian AI has first-class support for MCP, allowing easy integration with third-party services.
Does it support multi-agent teams?
Yes, you can create multi-agent teams, sequential pipelines, and parallel DAG workflows using a drag-and-drop canvas.