Peekaboo

New

Peekaboo is a macOS utility and MCP server that bridges visual screen content and AI assistants. It provides a high-performance CLI and an MCP-compatible server to capture lightning-fast screenshots, analyze images with cloud or local AI models, and perform full GUI automation on macOS. The project exposes services and tools that let AI agents and developers discover UI elements, manage windows and Spaces, extract menu hierarchies, and script interactions via natural language or JSON scripts. It is structured with shared core services (PeekabooCore), a Swift CLI, a native Mac app, and a TypeScript MCP server so agent integrations see consistent behavior. The README emphasizes privacy-friendly local inference via Ollama, multi-screen support, session persistence, and use by assistants like Claude Desktop and Cursor.

Stars

484

App URL

https://github.com/steipete/Peekaboo

Github Repository

https://github.com/steipete/Peekaboo/blob/main/README.md

Features

Fast, non-intrusive screenshot capture with per-window and multi-screen support. AI-powered image analysis using OpenAI, Anthropic, Grok, or local Ollama models. Full GUI automation tools including see (UI discovery), click, type, press, scroll, hotkey, swipe, move, drag and run for scripted workflows. Application, window, dock, menu, dialog and Space management with fuzzy matching and automatic focus management. Model Context Protocol (MCP) server and client integration with BrowserMCP and external tool support. Session management that preserves element IDs and state. Configurable logging, secure credential storage, performance optimizations for macOS APIs, and development helpers like Poltergeist auto-rebuild.

Use Cases

Peekaboo lets AI assistants and developers give visual context to models and automate macOS tasks without manual scripting of coordinates. It enables end-to-end automation workflows such as form filling, window arrangement across multiple displays, menu interaction, testing UI flows, and capturing verified screenshots for debugging. Agents can accept plain-language instructions which Peekaboo breaks into see-then-interact steps, reducing the effort to integrate vision into agent workflows. Local Ollama support provides a privacy-preserving option for on-device reasoning. The MCP server and CLI make it easy to wire Peekaboo into desktop assistants, IDE workflows, CI scripts, or larger multi-tool agent orchestrations.

Peekaboo

Basic Information

Links

App Details

Categories

Similar Listings

virtual lab

mnemo

EdgeChains

RAG Agents Accelerator

LLM Zero to Hundred

Featured Listings

Terry Bison Ranch

The Singapore Flyer

Tags

More Filters

Peekaboo

Categories

Similar Listings

virtual lab

mnemo

EdgeChains

RAG Agents Accelerator

LLM Zero to Hundred

Featured Listings

Terry Bison Ranch

The Singapore Flyer

Tags