GitHub - vercel-labs/agent-browser: Browser automation CLI for AI agents
Browser automation CLI for AI agents. Contribute to vercel-labs/agent-browser development by creating an account on GitHub.
Developer Tool
Summary
agent-desktop is a Rust CLI for AI agents to automate desktop apps via OS accessibility trees without screenshots. It uses progressive skeleton traversal to reduce tokens by 78-96% and offers FFI bindings for Python, Go, and Node.
Highlights
- Native Rust CLI with no runtime dependencies.
- Progressive skeleton traversal reduces token usage significantly.
- Deterministic element references for AI optimization.
- FFI bindings enable in-process calls from multiple languages.
- Supports any app with an accessibility tree.
auto-generated
Context
Audience
Software Developers and AI Engineers
DomainSoftware Engineering
Formatopen source software
Accessopen source
Source Overview
agent-desktop is a Rust-built CLI for AI agents to automate desktop apps via OS accessibility trees, outputting structured JSON with deterministic element references.
It features 53 commands for observation, interaction, keyboard/mouse control, and window management, using progressive skeleton traversal to reduce tokens by 78-96% on dense apps. Installation via npm downloads a prebuilt binary; FFI bindings enable in-process calls from Python, Swift, Go, Node, Ruby, and C.
Topics
Related
RustAI AgentsAccessibility APIsDesktop AutomationCLI ToolsFFI Bindings
