Skip to main content

​
What is OpenFang?

OpenFang is an open-source Agent Operating System — not a chatbot framework, not a Python wrapper around an LLM, not a “multi-agent orchestrator.” It is a full operating system for autonomous agents, built from scratch in Rust. Traditional agent frameworks wait for you to type something. OpenFang runs autonomous agents that work for you — on schedules, 24/7, building knowledge graphs, monitoring targets, generating leads, managing your social media, and reporting results to your dashboard. The entire system compiles to a single ~32MB binary. One install, one command, your agents are live.

​
Key Features

Autonomous Hands

7 pre-built capability packages that run independently on schedules: Clip (video processing), Lead (prospecting), Collector (OSINT), Predictor (forecasting), Researcher, Twitter, and Browser automation.

16 Security Layers

WASM dual-metered sandbox, Merkle hash-chain audit trail, taint tracking, Ed25519 signed manifests, SSRF protection, secret zeroization, and 10 more defense systems.

40 Channel Adapters

Connect agents to Telegram, Discord, Slack, WhatsApp, Signal, Matrix, Teams, LINE, Mastodon, Bluesky, and 30+ more messaging platforms.

27 LLM Providers

123+ models from Anthropic, OpenAI, Gemini, Groq, DeepSeek, xAI, Mistral, Ollama, and more with intelligent routing and automatic fallback.

​
Architecture

OpenFang is built from 14 Rust crates totaling 137,728 lines of code with a modular kernel design:
  • openfang-kernel — Orchestration, workflows, metering, RBAC, scheduler, budget tracking
  • openfang-runtime — Agent loop, 3 LLM drivers, 53 tools, WASM sandbox, MCP, A2A
  • openfang-api — 140+ REST/WS/SSE endpoints, OpenAI-compatible API, dashboard
  • openfang-channels — 40 messaging adapters with rate limiting, DM/group policies
  • openfang-memory — SQLite persistence, vector embeddings, canonical sessions, compaction
  • openfang-types — Core types, taint tracking, Ed25519 manifest signing, model catalog
  • openfang-skills — 60 bundled skills, SKILL.md parser, FangHub marketplace
  • openfang-hands — 7 autonomous Hands, HAND.toml parser, lifecycle management
  • openfang-extensions — 25 MCP templates, AES-256-GCM credential vault, OAuth2 PKCE
  • openfang-wire — OFP P2P protocol with HMAC-SHA256 mutual authentication
  • openfang-cli — CLI with daemon management, TUI dashboard, MCP server mode
  • openfang-desktop — Tauri 2.0 native app (system tray, notifications, global shortcuts)
  • openfang-migrate — OpenClaw, LangChain, AutoGPT migration engine

​
Why OpenFang?

OpenFang v0.1.0 is the first public release (February 2026). The architecture is solid and the test suite is comprehensive (1,767+ tests, 0 clippy warnings), but you may encounter breaking changes between minor versions until v1.0.

​
Performance Benchmarks

  • Cold start time: 180ms (vs 3-6s for Python frameworks)
  • Idle memory: 40MB (vs 180-400MB for competitors)
  • Install size: ~32MB single binary (vs 100-500MB for typical agent frameworks)

​
Security-First Design

OpenFang doesn’t bolt security on after the fact. Every layer is independently testable:
  • WASM dual-metered sandbox with fuel metering + epoch interruption
  • Merkle hash-chain audit trail (tamper-evident)
  • Information flow taint tracking for secrets
  • Ed25519 cryptographically signed agent manifests
  • SSRF protection blocking private IPs and cloud metadata endpoints
  • Secret zeroization (auto-wipes API keys from memory)

​
OpenAI-Compatible API

Drop-in replacement for existing tools: 
curl -X POST localhost:4200/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
    "model": "researcher",
    "messages": [{"role": "user", "content": "Analyze Q4 market trends"}],
    "stream": true
  }'

​
Next Steps

Quick Start

Get from install to your first working agent in under 5 minutes

Installation

Detailed installation instructions for all platforms and methods

Build docs developers (and LLMs) love

Get started for freeTalk to us