Skip to content

nullskymc/Aurevoy

Repository files navigation

Aurevoy

Aurevoy

Tell it what you want, not how to do it.
A personal AI agent that gets work done on your machine.

中文 · 한국어 · 日本語


Aurevoy is a local desktop agent for macOS, Windows, and Linux. You state a goal; it plans, uses tools, and works through the task—files, web research, code edits, and more—while you stay in control.

What you can do

"Convert PNGs in Downloads to WebP"
"Research React 19 breaking changes and write a migration checklist"
"Add dark mode to this project"
"Summarize these notes into a weekly report"
  • Code & files — search, precise edits, workspace-aware tools
  • Research — web search & fetch (SSRF-safe), optional browser skill
  • Multimodal — drag images/files; vision models when needed
  • Memory & knowledge — preferences you teach it; optional folder KB/RAG
  • Control — stop anytime; approve tools; inline edit & retry; branch threads

Everything runs on your computer. You bring your own API keys (OpenAI-compatible, Anthropic, and more). Open source.

Quick start

  1. Install — download the DMG (macOS), NSIS installer (Windows), or Linux package from Releases
  2. Configure — follow the in-app setup guide to connect a provider and select a model
  3. Chat — type a goal; / lists skills (e.g. web-search, browser)

Develop

# Node >= 22.19.0, Rust stable, macOS Xcode CLT
git clone https://github.com/nullskymc/Aurevoy.git
cd Aurevoy && npm install
npm run dev
npm run typecheck

Contributor guide: AGENTS.md · docs: docs/ · site: aurevoy.nullskymc.site

Stack

Layer Tech
Shell Tauri 2
UI React + TypeScript (packages/web-ui)
Engine Node + Fastify + Pi Agent + SQLite
Transport Local HTTP + SSE
Extensions Built-in tools + MCP + Skills

Architecture: docs/ARCHITECTURE.md · Roadmap: docs/ROADMAP.md

Capabilities (current)

Autonomous agent & tools · multi-provider LLM + OAuth · multi-turn & resume · inline edit-retry / branch / compact · multi-role subagents · skills · web search ·
multimodal · projects · dual task budgets · long-term memory + KB RAG · stdio + Streamable HTTP MCP · setup guide · tray + recent tasks · settings & CI

In progress: Apple code signing / notarization, implicit KB recall, evals, MCP presets, data-management polish. Shipped (distribution): macOS, Windows, and Linux packages plus in-app auto-update via GitHub Releases (Tauri updater).


MIT

About

Personal AI agent — tell it what you want, it figures out how. Runs locally, plans autonomously, writes code, browses the web, and gets things done.

Topics

Resources

Stars

Watchers

Forks

Releases

Packages

Contributors

Languages