OpenFang is an open-source Agent Operating System built from scratch in Rust, with more than 18,200 stars on GitHub. Instead of a chatbot framework that waits for prompts, OpenFang runs autonomous agents around the clock on schedules โ researching competitors, generating leads, monitoring targets, and reporting to a dashboard โ all from a single roughly 32 MB binary. Built by RightNow AI, it ships frequent releases and is under rapid development in 2026.
OpenFang’s core idea is Hands: pre-built autonomous capability packages that bundle a manifest, a detailed operational playbook, domain skills, and guardrails, all compiled into the binary. Seven Hands ship built in, covering research, lead generation, OSINT-style monitoring, forecasting, social media management, browser automation, and video clipping. Around them the kernel provides 40 channel adapters, 27 LLM providers with more than 120 models, MCP and A2A tool support, SQLite-plus-vector memory, and an OpenAI-compatible API. Security is a headline feature, with sixteen layered systems including a WASM sandbox, a Merkle hash-chain audit trail, secret zeroization, and mandatory approval gates for sensitive actions such as purchases.
OpenFang is completely free and open source under Apache-2.0 and MIT licensing. You pay only for the LLM provider usage your agents consume.
OpenFang is for builders who want always-on autonomous agents โ monitoring, research, lead generation, and content operations โ in one self-hosted binary instead of a tangle of frameworks.