RightNow-AI / openfang

OpenFang The Agent Operating System Open-source Agent OS built in Rust. 137K LOC. 14 crates. 1,767+ tests. Zero clippy warnings. One binary. Battle-tested. Agents that actually work for you. Documentation • Quick Start • Twitter / X --- > **v0.3.30 — Security Hardening Release (March 2026)** > > OpenFang is feature-complete but still pre-1.0. You may encounter rough edges or breaking changes between minor versions. We ship fast and fix fast. Pin to a specific commit for production use until v1.0. Report issues here. --- 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. Windows --- Hands: Agents That Actually Do Things "Traditional agents wait for you to type. Hands work for you." **Hands** are OpenFang's core innovation — pre-built autonomous capability packages that run independently, on schedules, without you having to prompt them. This is not a chatbot. This is an agent that wakes up at 6 AM, researches your competitors, builds a knowledge graph, scores the findings, and delivers a report to your Telegram before you've had coffee. Each Hand bundles: • **HAND.toml** — Manifest declaring tools, settings, requirements, and dashboard metrics • **System Prompt** — Multi-phase operational playbook (not a one-liner — these are 500+ word expert procedures) • **SKILL.md** — Domain expertise reference injected into context at runtime • **Guardrails** — Approval gates for sensitive actions (e.g. Browser Hand requires approval before any purchase) All compiled into the binary. No downloading, no pip install, no Docker pull. The 7 Bundled Hands | Hand | What It Actually Does | |------|----------------------| | **Clip** | Takes a YouTube URL, downloads it, identifies the best moments, cuts them into vertical shorts with captions and thumbnails, optionally adds AI voice-over, and publishes to Telegram and WhatsApp. 8-phase pipeline. FFmpeg + yt-dlp + 5 STT backends. | | **Lead** | Runs daily. Discovers prospects matching your ICP, enriches them with web research, scores 0-100, deduplicates against your existing database, and delivers qualified leads in CSV/JSON/Markdown. Builds ICP profiles over time. | | **Collector** | OSINT-grade intelligence. You give it a target (company, person, topic). It monitors continuously — change detection, sentiment tracking, knowledge graph construction, and critical alerts when something important shifts. | | **Predictor** | Superforecasting engine. Collects signals from multiple sources, builds calibrated reasoning chains, makes predictions with confidence intervals, and tracks its own accuracy using Brier scores. Has a contrarian mode that deliberately argues against consensus. | | **Researcher** | Deep autonomous researcher. Cross-references multiple sources, evaluates credibility using CRAAP criteria (Currency, Relevance, Authority, Accuracy, Purpose), generates cited reports with APA formatting, supports multiple languages. | | **Twitter** | Autonomous Twitter/X account manager. Creates content in 7 rotating formats, schedules posts for optimal engagement, responds to mentions, tracks performance metrics. Has an approval queue — nothing posts without your OK. | | **Browser** | Web automation agent. Navigates sites, fills forms, clicks buttons, handles multi-step workflows. Uses Playwright bridge with session persistence. **Mandatory purchase approval gate** — it will never spend your money without explicit confirmation. | **Build your own.** Define a with tools, settings, and a system prompt. Publish to FangHub. --- OpenFang vs The Landscape Benchmarks: Measured, Not Marketed All data from official documentation and public repositories — February 2026. Cold Start Time (lower is better) Idle Memory Usage (lower is better) Install Size (lower is better) Security Systems (higher is better) Channel Adapters (higher is better) LLM Providers (higher is better) Feature-by-Feature Comparison | Feature | OpenFang | OpenClaw | ZeroClaw | CrewAI | AutoGen | LangGraph | |---------|----------|----------|----------|--------|---------|-----------| | **Language** | **Rust** | TypeScript | **Rust** | Python | Python | Python | | **Autonomous Hands** | **7 built-in** | None | None | None | None | None | | **Security Layers** | **16 discrete** | 3 basic | 6 layers | 1 basic | Docker | AES enc. | | **Agent Sandbox** | **WASM dual-metered** | None | Allowlists | None | Docker | None | | **Channel Adapters** | **40** | 13 | 15 | 0 | 0 | 0 | | **Built-in Tools** | **53 + MCP + A2A** | 50+ | 12 | Plugins | MCP | LC tools | | **Memory** | **SQLite + vector** | File-based | SQLite FTS5 | 4-layer | External | Checkpoints | | **Desktop App** | **Tauri 2.0** | None | None | None | Studio | None | | **Audit Trail** | **Merkle hash-chain** | Logs | Logs | Tracing | Logs | Checkpoints | | **Cold Start** | ** = 18** installed (download) • OpenFang installed and initialized Setup **1. Install the gateway dependencies:** **2. Configure :** **3. Set the gateway URL (choose one):** Add to your shell profile for persistence: Or set it inline when starting the gateway: **4. Start the gateway:** The gateway listens on port by default. Override with . **5. Start OpenFang:** **6. Scan the QR code:** Open the dashboard → **Channels** → **WhatsApp**. A QR code will appear. Scan it with your phone: > **WhatsApp** → **Settings** → **Linked Devices** → **Link a Device** Once scanned…