Openfang, developed by Rightnow AI, is an open-source Agent Operating System built with Rust, designed for creating persistent autonomous agents. It installs as a single ~32MB binary with zero external dependencies, simplifying deployment across Linux, macOS, and Windows. This platform connects to numerous LLM providers and messaging platforms, offering a unified environment for agent operations. Openfang targets developers and teams focused on building production-ready AI agents, providing core capabilities and extensive integration options to manage complex, automated workflows. Its architecture supports peer-to-peer networking and various protocols for broad interoperability.

What does Openfang connect to?

Openfang acts as a central hub, integrating with 26-27 LLM providers, including Anthropic, OpenAI, Google Gemini, Groq, and DeepSeek. It supports 40 messaging platforms such as Telegram, Discord, Slack, WhatsApp, and Microsoft Teams. The system also functions as an MCP client, connecting to external Model Context Protocol servers like GitHub, filesystems, databases, and Puppeteer. For peer-to-peer agent networking, Openfang utilizes Google's Agent-to-Agent (A2A) protocol and its own OpenFang Protocol (OFP), enabling communication between other agents and Openfang instances.

Core features and tools

Openfang exposes 38 to 53+ built-in tools, including file_read and web_fetch, and can expose its own agents as MCP tools. It extends capabilities through SKILL.md-style skills, with 60 bundled. Core autonomous capabilities are packaged as "Hands": Clip, Lead, Collector, Predictor, Researcher, Twitter, and Browser. Openfang supports the Model Context Protocol (MCP) over HTTP and stdio, Google's A2A protocol, and its own OFP for P2P networking. It also offers a REST/WebSocket/SSE API with 76 endpoints and an OpenAI-compatible API.

Who is it for?

Openfang is intended for developers and teams actively building persistent autonomous agents for production environments. Its design supports complex, continuous operations rather than simple conversational interfaces or experimental prototypes. The platform's extensive integration capabilities with LLM providers, messaging channels, and external MCP servers make it suitable for advanced agent development. It caters to those needing to orchestrate multiple agent interactions and manage agent context across various systems. This is not the right fit if you are developing simple AI chat interfaces or prototypes.

How to get started

The server authenticates using an API key passed as an environment variable. Initial configuration requires setting a "SETUP_PASSWORD" for the web UI and an "OPENFANG_API_KEY" to secure API endpoints.

Use cases

• Lead Generation: The Lead Hand discovers, enriches, scores, and deduplicates qualified leads on a daily schedule, building ICP profiles and knowledge graphs.
• Competitive Intelligence/OSINT: Researcher or Collector Hands continuously monitor targets, detect changes, track sentiment, build knowledge graphs, and deliver cited reports.
• Social Media Management: The Twitter/X Hand manages autonomous posting schedules, creates content, responds to mentions, and tracks performance with an approval queue.

Openfang provides a specialized environment for developing and deploying production-grade autonomous agents. Its Rust foundation delivers a single, dependency-free binary, simplifying installation and operation. The platform's extensive protocol support and integrated "Hands" capabilities enable sophisticated agent workflows across diverse systems. Configuration is managed via a web dashboard, CLI, or files, with skills declaring required variables.