MCP Servers · openclaw

Peekaboo

Peekaboo is a macOS CLI and MCP server tool that lets AI agents capture screenshots and automate GUI interactions through natural language commands. It supports visual question answering via local or remote AI models and works with Claude, OpenAI, and Anthropic-compatible providers.

Talk to DEV.co Discuss Implementation

Source: GitHub — github.com/openclaw/Peekaboo

4.8k

GitHub stars

375

Forks

Swift

Primary language

MIT

License (OSI-approved)

Key facts

Objective fields from the source. Values we can't verify are shown as “Unknown” rather than guessed.

Field	Value
Repository	openclaw/Peekaboo
Owner	openclaw
Primary language	Swift
License	MIT — OSI-approved
Stars	4.8k
Forks	375
Open issues	8
Latest release	v3.7.1 (2026-07-05)
Last updated	2026-07-05
Source	https://github.com/openclaw/Peekaboo

What Peekaboo is

Written in Swift, Peekaboo provides pixel-accurate screen capture, accessibility-based UI automation (click, type, scroll, keyboard), menu/menubar discovery, and a native agent that chains commands. It runs as a standalone CLI, Homebrew app, or Node.js MCP server compatible with Claude, Codex, and Cursor.

Quickstart

Get the Peekaboo source

Clone the repository and explore it locally.

terminalbash

1git clone https://github.com/openclaw/Peekaboo.git2cd Peekaboo3# follow the project's README for install & configuration

Need it deployed, integrated, or customized instead? DEV.co ships production installs.

Best use cases

AI-Driven macOS Automation

Automate repetitive GUI tasks on macOS via natural-language agents without writing brittle coordinate-based scripts. Ideal for testing workflows, data entry, and application orchestration.

Headless Testing & Visual Regression

Capture multi-monitor setups at Retina resolution and analyze UI state programmatically. Supports CI/CD pipelines that need to verify macOS app behavior without human inspection.

MCP Integration for AI Code Assistants

Extend Claude, Cursor, or Codex with live macOS screen vision and click/type actions. Agents can see the current state and perform multi-step workflows within their development environment.

Implementation considerations

Requires explicit Screen Recording and Accessibility OS permissions; plan user onboarding and IT approval cycles. Background input delivery optional via --foreground flag for apps rejecting process-targeted input.
Snapshot-based state tracking; capture must precede element references. Plan for latency between `see` and `click` when UI changes rapidly.
AI provider configuration via environment variables (PEEKABOO_AI_PROVIDERS) or config file; test model temperature/max-token clamping for each provider.
Node 22+ required for MCP server; Swift 6.2 for CLI/app. Verify agent reliability in scripted workflows; natural language chains are non-deterministic.
Multi-screen support confirmed but test drag/window operations across screens; coordinate space handling edge cases should be verified in target setup.

When to avoid it — and what to weigh

Windows/Linux Primary Target — Peekaboo is macOS-only (requires macOS 15.0+/Sequoia). Not suitable for cross-platform automation or non-Apple workflows.
No System Permissions Available — Requires Screen Recording and Accessibility permissions; not deployable in locked-down MDM or kiosk environments without permission grants.
Headless Server Deployments — Designed for interactive machines with a display. Not intended for headless server automation or cloud-hosted agents that need macOS GUI access.
Production SLAs Without Vendor Support — Open-source community project; no commercial support tier or guaranteed SLA. Risky for critical business workflows requiring vendor responsibility.

License & commercial use

Licensed under MIT (permissive open-source). Permits commercial use, modification, and redistribution provided the license and copyright notice are retained.

MIT license permits commercial use without restriction. However, this is a community-maintained open-source project with no vendor support agreement, warranty, or SLA. Commercial deployments should assess build/support strategy independently and not assume vendor liability or patches.

DEV.co evaluation signals

Editorial assessment — not user reviews. Directional, with an explicit confidence level.

Signal	Assessment
Maintenance	Active
Documentation	Adequate
License clarity	Clear
Deployment complexity	Low
DEV.co fit	Good
Assessment confidence	High

Security considerations

Requires Screen Recording and Accessibility permissions—powerful OS capabilities. Review threat model if automating sensitive data input or screen capture. Background input delivery defaults to process-targeted (safer); foreground input visible to system. No mention of audit logging, clipboard encryption, or secrets handling. Daemon/socket model with permission isolation noted but security audit not provided in data. Evaluate for compliance requirements (e.g., FedRAMP, SOC 2) if used in regulated contexts.

Alternatives to consider

PyAutoGUI / PyRobotFramework

Cross-platform GUI automation; less AI-native but more mature, wider OS support, larger community. Steeper learning curve for ML agent integration.

OCR + Coordinate-Based Tools (Sikuli, ImageMagick + xdotool)

Lower-level, language-agnostic; no AI integration built-in. More portable across Unix-like systems but requires more glue code and OCR plumbing.

Apple's native Automation scripting (AppleScript, Shortcuts)

First-party, no third-party dependency; excellent for simple macOS workflows. No AI/vision capability, steeper learning curve for complex logic, less suited to LLM-based agents.

Software development agency

Build on Peekaboo with DEV.co software developers

Evaluate Peekaboo's fit for your workflow: test the Homebrew install, review permission requirements, and explore MCP integration with Claude or Cursor. Start with non-critical automation tasks to assess reliability.

Talk to DEV.co

Related open-source tools

Surfaced by semantic similarity across the DEV.co open-source index.

ghost-os browser-tools-mcp chrome-devtools-mcp

Related on DEV.co

Explore the category and the services that help you build with it.

MCP Servers AI Development Custom Software Talk to DEV.co

Peekaboo FAQ

Does Peekaboo work on non-macOS?

No. Peekaboo is macOS-only and requires macOS 15.0+ (Sequoia). Windows and Linux are not supported.

Can I use Peekaboo without granting Screen Recording and Accessibility permissions?

No. Peekaboo requires both permissions to function. There is no workaround for restricted/MDM environments.

What AI models are supported?

Any model exposed via Tachikoma multi-provider layer, including Claude (up to Fable 5 with 1M context), OpenAI-compatible, and Anthropic-compatible endpoints. Local LLM providers are also supported.

Is there a commercial support offering or SLA?

Unknown from provided data. Peekaboo is open-source MIT-licensed; no vendor support model is mentioned. Contact maintainer or community for support options.

Software development & web development with DEV.co

From first prototype to production, DEV.co delivers software development services around tools like Peekaboo. Our software development agency staffs experienced software developers and web developers for custom software development, web development, integrations, and ongoing support across mcp servers and beyond.

Ready to automate macOS with AI?

Talk to DEV.co Schedule a Consultation