ELVES is a Tauri-based desktop application designed for developers to orchestrate AI agent teams within isolated git worktree environments. It provides a neo-brutalist interface with embedded terminals, enabling users to type tasks in plain English and watch AI agents execute them in real-time.
Key Features
- Worktree Isolation: Each task runs in its own git worktree with a dedicated branch, preventing interference between parallel tasks.
- Embedded Terminal: A WebGL-accelerated PTY terminal connected to AI models like Claude Code or Codex, featuring real-time execution monitoring.
- Persistent Memory: SQLite-backed memory with FTS5 search that retains context, decisions, and preferences across sessions, with relevance decay over time.
- Task Analysis: Automatically classifies task complexity and selects appropriate modes (team or solo) before spawning agents.
- Runtime Agnostic: Seamlessly switch between different AI engines while maintaining the same interface and memory.
- Ship It Flow: One-click merge with strategy options (merge, rebase, squash), memory extraction, and worktree cleanup.
- Usage Insights: Dashboard with telemetry on token usage, daily activity, project breakdowns, and AI-generated narrative reports.
- Visual Editors: Catalog for skills, MCP servers, and project context with a split-pane layout for easy management.
Use Cases
- Software Development: Automate coding tasks like adding authentication, refactoring code, or implementing features.
- Project Management: Parallelize AI-driven tasks across multiple worktrees without conflicts.
- Code Review: Use AI agents to review and test changes in isolated environments before merging.
- Learning & Experimentation: Experiment with AI models in a controlled, persistent workspace.
Target Users
Developers, engineering teams, and tech-savvy individuals looking to integrate AI into their workflow efficiently, with a focus on macOS users initially (Linux and Windows support planned).
