One-binary runtime and desktop app
OpenFang ships as a Rust workspace with a single compiled binary, plus a native desktop app and a daemon for persistent operation.
OpenFang
ClaimOpenFang is an open-source Agent OS in Rust for running, configuring, and supervising autonomous agents locally or self-hosted.
Overview
OpenFang is an open-source Agent OS built in Rust for installing, configuring, and operating autonomous agents. The public docs position it as a runtime for hands, agents, tools, security, memory, channels, protocols, and a native desktop app, with a daemon for persistent and multi-user use.
It is aimed at people who want to run agents locally or self-host them, choose an LLM provider, attach messaging channels, and manage workflows from templates or custom manifests. The docs also show support for browser-based chat, API access, MCP, A2A, and a native desktop application.
Core capabilities
Pre-built autonomous Hands
The product includes 7 pre-built autonomous Hands such as Clip, Lead, Collector, Predictor, Researcher, Twitter, and Browser, each meant for a specific workflow.
Security hardening stack
The docs describe 16 security systems including WASM dual metering, manifest signing, audit trails, taint tracking, SSRF protection, secret zeroization, mutual auth, and prompt-injection scanning.
Channel, model, and skill integration
OpenFang supports 40 messaging channel adapters, 20 LLM providers, and 60 bundled skills, with additional agent templates and workflow composition.
Multiple deployment and setup paths
The getting started guide shows installation via shell script, Cargo, or Docker, then configuration through `config.toml` and environment variables.
API and protocol support
The docs expose REST, WebSocket, and SSE endpoints, plus an OpenAI-compatible `/v1/chat/completions` API for external integration.
Practical uses
Start with a pre-built agent
Install OpenFang, set an API key, run `openfang init`, and spawn a built-in template such as `hello-world` to get a working agent running quickly.
Define custom agents
Create a custom `agent.toml` with a chosen provider, model, and capability set when you want an agent tailored to a specific task or workflow.
Operate agents continuously
Run `openfang start` to expose the REST API, WebSocket endpoint, WebChat UI, and persistent daemon behavior for ongoing use.
Attach external systems
Connect messaging platforms, MCP servers, or A2A integrations when you need agents to interact with external tools and channels.
Manage agents interactively
Use the native desktop app or browser chat interface when you want a local UI for monitoring running agents, reading responses, and checking token usage.
Pros and Cons
Pros
- Open-source and built in Rust, with installation paths for shell, Cargo, and Docker.
- Provides a broad agent runtime surface: templates, custom manifests, daemon mode, desktop app, and browser chat.
- Includes documented security features such as sandboxing, audit trails, taint tracking, secret zeroization, and prompt-injection scanning.
- Supports multiple integration surfaces, including messaging channels, skills, MCP, A2A, REST, WebSocket, and SSE.
- The getting started guide gives concrete setup steps, from API keys to `openfang doctor` and agent spawning.
Cons
- The public pricing page is unavailable in the collected sources, so pricing and commercial terms are not confirmed here.
- The docs shown here are rich on setup and architecture, but not all provider, channel, and workflow details are fully enumerated on the pages used for this extract.
FAQ
How do I install OpenFang?
OpenFang’s getting started guide shows three installation paths: a shell installer for Linux and macOS, `cargo install` for any platform with Rust 1.75+, and Docker.
What do I need to configure before running agents?
The docs say OpenFang needs at least one LLM provider API key. It supports environment variables such as `ANTHROPIC_API_KEY`, `OPENAI_API_KEY`, and `GROQ_API_KEY`.
Can I use OpenFang with built-in templates or custom agents?
The getting started guide shows both built-in agent templates and custom `agent.toml` manifests. You can spawn a template agent or define your own model and capability settings.
What does the daemon add beyond the CLI?
The daemon supports persistent agents, multi-user access, the WebChat UI, REST API, WebSocket access, and token usage visibility in the browser UI.
What are OpenFang’s main limitations from the public docs?
The architecture and docs pages describe OpenFang as a Rust workspace with sandboxing, security hardening, channel adapters, skills, MCP, and A2A support. The public docs do not provide a complete list of every supported provider or integration detail on the pages used here.
Quick Facts
- Category
- Developer Tool
- Product type
- Open-source agent operating system
- Platform
- Linux, macOS, Windows
- Source domain
- openfang.sh
- Primary users
- Developers and teams self-hosting agent workflows
- Deployment
- CLI, daemon, native desktop app, Docker
OpenFang Alternatives
Kastra
Kastra authorization infrastructure for AI systems checks prompts, tool calls, shell commands, API requests, and browser actions before execution.
Orca
Orca is an Agent Development Environment for shipping with coding agents, running multiple CLI agents in parallel across isolated worktrees, with desktop and mobile workflows.
AI Magicx
AI Magicx is a unified AI workspace for chat, image, video, voice, music, email and developer tasks, helping teams and creators manage multiple models in one place.
Paper
Paper is a design tool that connects canvas, code, and AI agents so teams can create, share, and ship work in one workflow. Includes desktop app and MCP access.
blop
blop is a QA agent that writes browser tests as code in your repo, runs them in CI, clusters repeated failures, and can open PRs to fix broken tests.
RLAMA
RLAMA is a local AI platform for building RAG systems and intelligent agents on macOS, Linux, and Windows, with HTTP API support.