RightNow-AI/openfang

⭐ 17,330  ·  Rust  ·  GitHub Repo

Open-source Agent Operating System

agent-framework ai-agents llm mcp open-source openclaw operating-system rust

1-Sentence Summary

Rust-built Agent OS running autonomous AI workers 24/7 on schedules, not a chatbot framework.

🔥 Key Capabilities & USP

• 7 Autonomous "Hands" (Capability Packages): Pre-built agents for lead generation, research, social media, content creation, browser automation, and prediction. They run on schedules without requiring any user prompts—set them and walk away. Solves the "always-on, always-working" automation gap that chatbot frameworks leave unfilled.
• 16 Security Systems (Defense-in-Depth): WASM dual-metered sandbox, Merkle hash-chain audit trail, and mandatory approval gates for sensitive actions. This isn't a toy—it's production-grade isolation that lets you trust autonomous agents with real tasks. Solves the "how do I let an agent touch my data without fear" problem.
• 40 Channel Adapters: Plug into Slack, Discord, email, Telegram, webhooks, databases, and more. Your agents can output to any channel and ingest from any source. Solves the integration fragmentation that kills most agent projects.
• Single ~32MB Binary Deployment: One command (openfang start) launches all agents. No Docker, no Python venvs, no dependency hell. Solves the "it works on my machine" deployment nightmare.
• MCP & A2A Protocol Support: 53 built-in tools plus standard protocol compatibility. Your agents aren't locked into a proprietary ecosystem—they speak the emerging standards.

Technical Architecture

Component	Detail
Language	Rust (137K LOC across 14 crates)
Agent Runtime	WASM-based sandbox with dual-metered resource limits (CPU + memory)
Storage	SQLite + vector memory for hybrid retrieval
Desktop UI	Tauri 2.0 dashboard at localhost:4200
Binary Size	~32MB single executable
Performance	Cold start <200ms, idle memory ~40MB
Testing	2,543+ passing tests
Security	Merkle hash-chain audit trail, approval gates, WASM sandbox

Quick Start Guide

# Install (Linux/macOS)
curl -fsSL https://openfang.sh/install | sh

# Initialize and start all agents
openfang init
openfang start

# Dashboard live at http://localhost:4200

# Install (Windows)
irm https://openfang.sh/install.ps1 | iex
openfang init
openfang start

# Activate the Researcher Hand. It starts working immediately.
openfang hand activate researcher

# Check its progress anytime
openfang hand status researcher

# Activate lead generation on a daily schedule
openfang hand activate lead

# Pause without losing state
openfang hand pause lead

# See all available Hands
openfang hand list

Pros, Cons & Use Cases

Pros

• Autonomous by default: No prompting required—agents run on cron-like schedules. This is the killer feature that separates it from every chatbot framework.
• Extreme performance: 200ms cold start and 40MB idle memory means this runs on a Raspberry Pi or a $5 VPS.
• Security-first architecture: 16-layer defense system with audit trails is rare in the agent space. You can actually trust these agents with real work.
• Single binary deployment: No dependency management, no container orchestration—just a curl and a command.
• Rust foundation: Memory safety, no garbage collector pauses, and excellent concurrency for running multiple agents simultaneously.

Cons

• Pre-1.0 software: Breaking changes between minor versions. You must pin to specific commits for production deployments.
• Rough edges expected: As with any early-stage project, documentation gaps and unexpected behavior are likely.
• Rust learning curve: If you need to extend the core or write custom Hands, you're writing Rust—not Python or TypeScript.

Who should NOT use this?

• Teams needing a chatbot for customer support: This is not a conversational AI framework. It's an autonomous agent OS. If you want a chatbot, look elsewhere.
• Production environments requiring API stability: Until v1.0, expect breaking changes. If you can't tolerate version pinning and manual upgrades, wait.
• Non-developer users: The CLI and dashboard assume technical proficiency. This is not a no-code tool.
• Teams needing extensive custom agent logic without Rust: If your team only knows Python/JS, writing custom Hands will be a significant barrier.

Ideal Use Cases

• Automated lead generation: Activate the Lead Hand on a daily schedule to scrape, enrich, and qualify leads while you sleep.
• Social media management: The Twitter Hand posts on schedule, the Browser Hand monitors trends—fully autonomous.
• Competitive intelligence: The Researcher Hand runs nightly reports on competitors, the Predictor Hand forecasts market moves.
• OSINT intelligence gathering: Scheduled, auditable, and sandboxed—perfect for security researchers who need repeatable, trustworthy data collection.
• Content pipelines: The Clip Hand curates, the Collector Hand aggregates, and the Browser Hand publishes—all on cron schedules.

Community & Activity

OpenFang is on fire. With 17,330 stars and active development through May 2026, this project has captured the imagination of the autonomous agent community. The Rust ecosystem is rallying behind it—2,543+ tests, 137K LOC across 14 crates, and a clean CLI experience signal serious engineering discipline. The team is shipping fast (pre-1.0 with breaking changes means rapid iteration), and the Tauri 2.0 desktop app shows they're thinking about developer experience, not just backend plumbing. If you want to be part of the next wave of agent infrastructure—before it hits v1.0 and the API locks down—now is the time to jump in.