OpenFang — The Open-Source AI Agent Operating SystemOpenFang

Open-source AI agent framework built in Rust. Deploy 7 autonomous Hands that run on schedules, build knowledge graphs, and report to your dashboard — single binary, zero dependencies.

14 crates · 137K lines of Rust · Zero clippy warnings · Battle-tested architecture

$ curl -fsSL https://openfang.sh/install | sh
✍️
Writer
bot

Write a blog post about AI agents in 2026 — SEO-optimized, with sources.

0:02✓✓

Writer is composing...

Published! 2,400-word article covering autonomous agents, multi-model orchestration, and enterprise adoption. SEO score 94/100 with 8 cited sources.

📝 2,400 words
🔍 SEO score: 94/100
📚 8 sources cited
🏷️ Meta tags optimized
0:02
40+
Channels
26+
Providers
16
Security layers
7
Autonomous Hands
MIT + Apache-2.0·License

AI Agent Framework Capabilities

Everything you need to run autonomous AI agents in production.

Autonomy

7 Autonomous Hands

Pre-built agents: Clip, Lead, Collector, Predictor, Researcher, Twitter, Browser. Run on schedules, build knowledge graphs, report to dashboard.

Security

16 Security Systems

WASM dual-metered sandbox, Ed25519 manifest signing, Merkle audit trail, taint tracking, SSRF protection, secret zeroization, and more.

Isolation

Sandboxed Execution

WASM tool execution with dual metering. Workspace-confined file operations. Environment-cleared subprocess isolation. 10-phase graceful shutdown.

Agents

30 Pre-Built Agents

Four performance tiers across Anthropic, Gemini, Groq, and DeepSeek. Ready to deploy out of the box.

Tools

38 Built-In Tools + MCP

38 native tools plus Model Context Protocol client and server. Connect external MCP servers and expose OpenFang tools to other agents.

Memory

Persistent Memory

SQLite-backed storage with vector embeddings. Cross-channel canonical sessions, automatic LLM-based compaction, and JSONL session mirroring.

Channels

40 Channel Adapters

Telegram, Discord, Slack, WhatsApp, Teams, IRC, Matrix, and 33 additional platforms. Connect your agents everywhere.

Protocols

MCP + A2A + OFP

Model Context Protocol, Google A2A agent-to-agent tasks, OpenFang Protocol for P2P networking between instances.

Desktop

Tauri 2.0 Native App

Desktop application with system tray, notifications, single-instance enforcement, auto-start, and global shortcuts.

OpenFang vs Other AI Agent Frameworks

Compared across 10 dimensions against leading frameworks.

DimensionOpenFangOpenClawZeroClawCrewAIAutoGenLangGraph
LanguageRustTypeScriptRustPythonPythonPython
Autonomous Hands7 built-inNoneNoneNoneNoneNone
Security Layers1636121
Agent SandboxWASM dualDockerWASMNoneDockerNone
Channel Adapters40815000
Built-in Tools53+MCP3050+122015
MemorySQLite+vecExternalFile-basedExternalExternalCheckpoints
Desktop AppTauri 2.0NoneNoneNoneStudioNone
Audit TrailMerkle chainLogsLogsLogsCheckpointsLogs
LicenseMITMITMITMITMITMIT

Performance Benchmarks

Real benchmarks across key performance metrics. Measured, not marketed.

Cold Start

ZeroClaw
10ms
OpenFang
180ms
LangGraph
2500ms

Idle Memory

ZeroClaw
5MB
OpenFang
40MB
OpenClaw
394MB

Install Size

ZeroClaw
8.8MB
OpenFang
32MB
OpenClaw
500MB

Security Systems

OpenFang
16
ZeroClaw
6
CrewAI
1

Channel Adapters

OpenFang
40
ZeroClaw
15
OpenClaw
8

LLM Providers

ZeroClaw
28
OpenFang
27
OpenClaw
12

Autonomous AI Agents That Run on Schedules

Unlike regular agents you chat with, Hands run on schedules, build knowledge graphs, and report results to your dashboard.

Clip

Content

Video-to-shorts converter with captions and thumbnails. 8-phase pipeline powered by FFmpeg and yt-dlp.

8-phase pipelineFFmpeg + yt-dlp5 STT backendsAuto-publish

Lead

Data

Daily lead generation with ICP scoring. Web research loops with CSV/JSON/Markdown export.

ICP scoring 0-100Web research loopsCSV/JSON/MD export

Collector

Intelligence

OSINT monitoring with sentiment tracking. Change detection and knowledge graph construction.

Change detectionSentiment analysisKnowledge graphsAlerts

Predictor

Forecasting

Superforecasting with Brier scores. Calibrated probability estimates with contrarian mode.

Brier score calibrationContrarian modeEvidence chains

Researcher

Productivity

CRAAP fact-checking with multi-language support. APA citations and structured reports.

CRAAP fact-checkingMulti-languageAPA citations

Twitter

Communication

X account management with 7 content formats. Approval queue and engagement tracking.

Approval queue7 content typesEngagement trackingBrand voice

Browser

Automation

Web automation with purchase approval gates. Playwright bridge with session persistence.

Purchase gatePlaywright bridgeSession persistenceCAPTCHA detection

Hand Commands

openfang hand --help
openfang hand activate  <hand>
openfang hand deactivate <hand>
openfang hand status    <hand>
openfang hand list
openfang hand pause     <hand>
openfang hand resume    <hand>

Documentation & Quick-Start Guides

Comprehensive documentation to get you started and keep you going.

01

Quick Start Guide

Installation and first agent deployment

02

Architecture Overview

14-crate Rust workspace, kernel design

03

Configuration Reference

config.toml fields, 62 environment variables

04

Security Model

16 security systems: WASM sandbox, taint tracking, audit trail

05

LLM Providers

26 providers, 50+ models, cost rates, routing

06

40 Channel Adapters

Platform integration guide

07

Agent Templates

30 pre-built agents across 4 performance tiers

08

Hands System

Configuration and deployment guide

09

Skill Development

60 bundled skills, SKILL.md format

10

Workflow Engine

Multi-agent pipelines: fan-out, conditional, loop modes

11

MCP & A2A Protocols

Inter-agent communication reference

12

REST API Reference

140+ endpoints: agents, memory, workflows, channels

Frequently Asked Questions

Common questions about OpenFang.

What is OpenFang?
OpenFang is an open-source Agent Operating System built from scratch in Rust. It is not a chatbot framework or a Python wrapper around an LLM — it is a full operating system for autonomous agents, compiling to a single ~32MB binary with 7 bundled autonomous Hands, 40 channel adapters, 38 built-in tools, and 16 security systems.
What are Hands?
Hands are pre-built autonomous capability packages. Unlike regular agents you chat with, Hands run on schedules, build knowledge graphs, and report results to your dashboard. OpenFang ships with 7 Hands: Clip (content), Lead (data), Collector (intelligence), Predictor (forecasting), Researcher (productivity), Twitter (communication), and Browser (automation).
How does OpenFang handle security?
OpenFang features 16 security systems including WASM dual-metered sandbox, Ed25519 manifest signing, Merkle hash-chain audit trail, taint tracking, SSRF protection, secret zeroization, HMAC-SHA256 mutual auth, GCRA rate limiter, subprocess isolation, prompt injection scanner, and path traversal prevention.
What LLM providers does OpenFang support?
OpenFang supports 26 LLM providers with 50+ models, including Anthropic, OpenAI, Google Gemini, Groq, DeepSeek, and more. It includes an OpenAI-compatible API and intelligent routing across providers.
How do I install OpenFang?
Install with a single command: curl -fsSL https://openfang.sh/install | sh. OpenFang supports macOS, Linux, and Windows. The entire system compiles to a single ~32MB binary with zero external dependencies.
What platforms can agents connect to?
OpenFang includes 40 channel adapters: Telegram, Discord, Slack, WhatsApp, Microsoft Teams, IRC, Matrix, and 33 additional platforms. Agents can operate across multiple channels simultaneously with cross-channel canonical sessions.
Is OpenFang open source?
Yes. OpenFang is fully open source under the MIT license. The complete codebase is available on GitHub at github.com/RightNow-AI/openfang.
How does OpenFang compare to other AI agent frameworks?
OpenFang is built in Rust (not Python), offers 16 security layers (vs 1-2), includes 40 channel adapters (vs 0), ships with 7 autonomous Hands (unique to OpenFang), provides a native Tauri 2.0 desktop app, and features a Merkle chain audit trail. It cold-starts in 180ms vs 2500ms for LangGraph.
Built & maintained by

RightNow AI

OpenFang is built by Jaber, Founder of RightNow AI. Passionate about building reliable, secure, and performant AI infrastructure for production deployments.

rightnowai.co@Akashi203