Overview
OpenFang describes itself as an agent operating system rather than a chatbot framework: its unit of work is an agent that runs on a schedule, unattended, instead of one that waits for a prompt. The whole system is written in Rust and compiles to a single binary of roughly 32 MB with no external dependencies, so a deployment is one file plus a config.
Capabilities ship as pre-built packages the project calls "hands" — among them Clip for video processing, Lead for prospect generation, Collector for OSINT, Predictor for forecasting, Researcher for deep analysis, Twitter for account management, and Browser for web automation. You activate the hands an agent needs rather than assembling tools yourself.
The codebase spans 14 Rust crates covering kernel orchestration, a runtime with 53 tools, 140+ REST/WebSocket/SSE endpoints, 40 channel adapters and a SQLite memory layer, plus a native desktop app. Security is a stated design axis: a dual-metered WASM sandbox, a Merkle hash-chain audit trail, Ed25519 manifest signing, SSRF protection and secret zeroization are among the 16 independent systems the project enumerates. It connects to 27 LLM providers and 40 messaging channels including Telegram, Discord, Slack and WhatsApp.
What it does
- Scheduled, long-running autonomous agents rather than request-response chat sessions
- Ships as one ~32 MB dependency-free Rust binary, with a native desktop app alongside
- Pre-built "hands" — Researcher, Collector, Predictor, Browser, Clip, Lead, Twitter — activated per agent
- WASM dual-metered sandbox, Merkle hash-chain audit trail and Ed25519 manifest signing
- 27 LLM providers (Anthropic, OpenAI, Gemini and others) and 40 messaging channel adapters
- 140+ REST, WebSocket and SSE endpoints plus a SQLite-backed memory layer
Getting started
OpenFang installs from a single shell script and configures its LLM provider interactively. A local dashboard comes up on port 4200.
Install the binary
macOS and Linux use the shell installer; Windows uses the PowerShell equivalent.
curl -fsSL https://openfang.sh/install | shInstall on Windows
The PowerShell one-liner does the same thing.
irm https://openfang.sh/install.ps1 | iexInitialize and start
openfang init walks through LLM provider configuration interactively; openfang start brings up the runtime and the dashboard at http://localhost:4200.
openfang init
openfang startActivate a hand and talk to it
Hands are the capability packages. Activate one, then chat with the agent it backs.
openfang hand activate researcher
openfang chat researcherOptional: WhatsApp Web channel
The WhatsApp Web adapter needs Node.js 18+ and a gateway URL in the environment.
export WHATSAPP_WEB_GATEWAY_URL="http://127.0.0.1:3009"Commands and code are distilled from the project's own documentation — always check the official repo for the latest.
When to use it
- Run a research or monitoring agent on a fixed schedule and have it report into Slack, Telegram or Discord
- Deploy agents where a single self-contained binary is easier to ship and audit than a Python stack
- Sandbox untrusted agent capabilities in WASM with a tamper-evident audit trail of what ran
- Automate outbound or OSINT collection workflows with pre-built capability packages instead of custom tool code
How OpenFang compares
OpenFang alongside other open-source agent frameworks & builders tools AI/TLDR tracks, ranked by GitHub stars.
| Tool | Stars | What it does |
|---|---|---|
| DeepSeek Harness | ★ 226k | DeepSeek AI's open-source agent harness (dsh), built on Cordis, where models, tools, skills, sessions, sandboxes, storage and the UI are all plugins composed through profiles. |
| AutoGPT | ★ 187k | One of the earliest autonomous agent projects, now a platform for building and running agents from reusable blocks and workflows. |
| DeerFlow | ★ 82.5k | ByteDance's open-source super agent harness built on LangGraph: skills, sub-agents, sandboxes, a filesystem and long-term memory for long-horizon research, coding and content tasks. |
| nanobot | ★ 48.2k | Lightweight self-hosted personal AI agent framework in Python, with a WebUI, terminal and chat-app channels, tools, long-term memory, MCP and scheduled automations. |
| Agno | ★ 42.2k | A fast Python framework (formerly Phidata) for building agents with memory, tools, and multimodal inputs, plus a runtime for deploying them in production. |
| LangGraph | ★ 41.8k | A library from the LangChain team for building stateful, graph-based agent workflows with explicit control over steps, memory, and human-in-the-loop checkpoints. |
| AgentGPT | ★ 36.3k | AgentGPT lets you name a custom AI, give it a goal, and watch it plan tasks, run them, and learn from the results, all from a web browser. |
| OpenFang | ★ 18.2k | Single-binary Rust runtime for agents that run on a schedule |