osaurus
Own your AI. The open-source agent harness for macOS. Run local or cloud models with persistent memory, skills, tools, and autonomous execution. Built in Swift.
What is osaurus?
Own your AI. The open-source agent harness for macOS. Run local or cloud models with persistent memory, skills, tools, and autonomous execution. Built in Swift.
Overview of osaurus
What it is
Osaurus is a native macOS harness for AI agents, built purely in Swift. It sits between the user and any local or cloud model, providing persistent memory, autonomous execution, and cryptographic identity. It works fully offline with local models and can connect to cloud providers. The project is open source and MIT licensed.
Key features
- Agents: Each agent has its own prompts, memory, and visual theme. Tools and skills are selected via RAG search based on the task.
- Sandbox: Agents execute code in an isolated Linux VM powered by Apple's Containerization framework, with full dev environment (shell, Python, Node.js, compilers, package managers). Falls back to a macOS Seatbelt sandbox on macOS versions prior to Tahoe.
- Memory: Three layers — identity, pinned facts, and per-session episodes — plus a transcript fallback. Agents distill conversations once at session end and surface at most one compact slice per request.
- Privacy Filter: On-device classifier detects and scrubs sensitive data before sending to cloud models; streaming replies are unscrubbed so the chat reads naturally. Fail-closed: if the post-scrub scan finds anything leaked, the send is blocked.
- Identity: Every participant gets a secp256k1 cryptographic address. Authority flows from a master key (iCloud Keychain) down to each agent in a verifiable chain of trust.
- Relay: Expose agents to the internet via secure WebSocket tunnels through agent.osaurus.ai, with unique URL per agent based on its crypto address.
- Secure Channel: End-to-end encrypted conversations between agents using forward-secret X25519 handshake and ChaCha20-Poly1305 encryption.
Use cases
Research assistant, coding partner, file organizer, or any other custom agent. Users can spawn custom agents and delegate work to subagents in parallel, within set budgets. The built-in Orchestrator answers questions about the app and can change settings after user approval.
Getting started
bash brew install --cask osaurus After installing, launch from Spotlight or the CLI: bash osaurus ui # Open the chat UI osaurus serve # Start the server osaurus status # Check status Requires macOS 15.5+ and Apple Silicon.
License and pricing
MIT license. Open source. Pricing type: open_source.
Based on the project's public README and repository metadata.