# osaurus-ai/osaurus

Own your AI. The native macOS harness for AI agents -- any model, persistent memory, autonomous execution, cryptographic identity. Built in Swift. Fully offline. Open source.

Repository: https://github.com/osaurus-ai/osaurus
Canonical: https://ross.abutalabs.com/products/osaurus
Homepage: https://osaurus.ai
Language: Swift
License: MIT
License Family: permissive
Topics: llm, swift, apple-foundation-models, apple-intelligence, apple-neural-engine, mlx, mcp, mcp-server, anthropic, openai
Last push: 2026-08-26T22:53:18+00:00

## Health v2 (maintenance only)
Score: 80/100 (v2, computed 2026-09-03T02:39:23.370411+00:00)
- activity 99, release rhythm 87, longevity 27
- inputs: {"age_days": 381, "days_push": 7, "days_rel": 10, "gap_med": 0, "n_releases_24m": 454}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 7691, forks 453 (observed 2026-08-28T04:10:02.451716+00:00)

## What it is
Osaurus is a native Swift macOS application that acts as a local-first AI harness, running open models on Apple Silicon (via MLX and Apple Foundation Models) with support for cloud providers like ChatGPT, Claude, and Gemini. It provides autonomous agents with persistent memory, tool execution, cryptographic identity, and OpenAI-, Anthropic-, and Ollama-compatible local API servers plus MCP integration, all fully offline-capable and MIT licensed.

## Use cases
- run local LLMs on my Mac with full privacy
- build autonomous AI agents that work offline
- serve an OpenAI-compatible API from local models
- connect AI agents to tools via MCP
- keep a shared memory across ChatGPT, Claude, and local models
- run AI inference on Apple Neural Engine
- self-host a private AI assistant with no telemetry

## When to choose
- you want a fully offline, private AI assistant on Apple Silicon
- you need a native macOS app (no Electron) with local model inference
- you want agents with persistent memory and autonomous execution under your control
- you need a local OpenAI/Anthropic/Ollama-compatible server or MCP endpoint

## When to avoid
- you need Linux or Windows support - it is macOS Apple Silicon only
- you need maximum model breadth - tools like Ollama or LM Studio support more architectures
- you want a headless server-only inference stack rather than a desktop app

## Facets
- artifact type: application
- maturity: active
- function: llm-inference, agent-framework, mcp, chatbot, rag, http-server, cli, plugin-system
- domain: artificial-intelligence, large-language-models, chatbots, privacy, self-hosted, developer-tools
- platform: -
- tags: local-ai, apple-silicon, mlx, apple-foundation-models, offline-ai, openai-compatible-api, anthropic-compatible-api, ollama-compatible, autonomous-agents, persistent-memory, native-swift, ai-agents, macos, desktop, swift

## Member repositories
- osaurus-ai/osaurus (main) score 80

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:10:02.451716+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T17:36:01.010922+00:00, confidence not recorded.
  - readme: https://github.com/osaurus-ai/osaurus (fetched 2026-08-28T04:10:02.451716+00:00, sha dc180362d1d3)
  - homepage: https://osaurus.ai (fetched 2026-08-29T08:31:36.846728+00:00, sha 9512f6e9b359)
  - site_page: https://osaurus.ai/about (fetched 2026-08-29T08:31:36.852462+00:00, sha 89b408713154)
  - site_page: https://docs.osaurus.ai/ (fetched 2026-08-29T08:31:36.854986+00:00, sha f003f9bea406)
  - site_page: https://osaurus.ai/changelog (fetched 2026-08-29T08:31:36.857176+00:00, sha 363b014a1ce0)
- Data as of 2026-08-30T08:39:29.467469+00:00.