Claude Marketplaces
SkillsMCP serversMarketplacesDigestTools
Submit

This week in Claude

Every Monday: Claude Code, Agent SDK, MCP, and the Anthropic platform moves worth your time.

Skills by category
Frontend DevelopmentBackend & APIsTesting & QASecurityDevOps & CI/CDGit & Pull RequestsDocumentationCode Review & QualityAI & Agent BuildingSkill Development
MCP servers by category
Sales & MarketingWeb & Browser AutomationDatabasesAI & LLM ToolsCloud & InfrastructureCommunication & MessagingDeveloper ToolsDesign & CreativeDocuments & KnowledgeSearch & Web Crawling
Marketplaces by category
AI Agents & OrchestrationLLM IntegrationDevelopment ToolsFrontend & UIBackend & APIsDatabasesTesting & Code QualityDevOps & CloudSecurity & ComplianceGit & Version Control
Claude Marketplaces

An independent directory of skills, plugins and MCP servers for Claude Code. Not affiliated with Anthropic.

Browse

  • Skills
  • MCP servers
  • Marketplaces
  • Weekly digest
  • Skill index
  • MCP index
  • Marketplace index

Site

  • About
  • Tools
  • Submit
  • Feedback
  • Privacy
  • llms.txt
  • Plugins reference

Maker

  • Snug
  • Agent Caffeine
  • knobs
  • Wax
  • Vectoys
  • mertduzgun.com
Crafted at
Skills
steipete avatar

Peekaboo

steipete/clawdis

This is a comprehensive macOS UI automation CLI that gives Claude direct control over your desktop. Instead of asking you to click things, Claude can capture annotated screenshots with element IDs, then click, type, drag, and navigate menus itself. The see-then-click workflow is notably practical: capture an annotated view, get element IDs, then target them directly. It covers the full scope from basic screenshots to window management, menu navigation, and even live motion-aware capture. Requires Screen Recording and Accessibility permissions, obviously. If you're building agents that need to interact with native Mac apps or automate complex UI workflows, this is the missing piece between Claude's reasoning and actual desktop control.

installs
3.4K
GitHub stars
382.8K
last updated
Jul 14, 2026
View on GitHubSave

Install

$ npx -y skills add steipete/clawdis --skill peekaboo --agent claude-code

Installs into .claude/skills of the current project.

Files

Peekaboo

Use Peekaboo to inspect macOS UI, act on the intended target, and verify the result. The examples below use v4 syntax. Check peekaboo --version and the installed command's --help; for older versions, follow that version's help.

OpenClaw Bridge

The OpenClaw macOS app hosts Peekaboo Bridge when Computer Control is enabled, its provider is Peekaboo, and Peekaboo Bridge is enabled. Keep the existing OpenClaw socket selection when running through that host:

export PEEKABOO_BRIDGE_SOCKET="${PEEKABOO_BRIDGE_SOCKET:-$HOME/Library/Application Support/OpenClaw/bridge.sock}"
peekaboo bridge status --json

Confirm that the selected host and socket are the intended ones. Preserve an explicit PEEKABOO_BRIDGE_SOCKET or --bridge-socket selection. If the required host is unavailable, report that outcome; do not clear the selection or switch Computer Control providers to make a command succeed.

Permissions belong to the process or Bridge host performing the operation. Do not pass --no-remote unless the caller has the required permissions.

Find the installed command

Use leaf help for flags and targeting requirements instead of copying a flag between commands. Use root help to discover the available commands:

peekaboo --help
peekaboo app list --help
peekaboo see --help
peekaboo press --help
peekaboo drag --help

V4 removed these old command roots:

Old commandV4 command
list appsapp list
imagesee --no-elements for pixels without element detection
hotkeypress with chords such as cmd+shift+t
swipedrag --from … --to …

For other operations, including clicking, typing, scrolling, and managing windows or menus, use peekaboo <command> --help. Put command options after the leaf command; --json requests structured output.

Inspect, act, verify

Start with application inventory and a fresh view of the target:

peekaboo app list --json
peekaboo see --app Safari --window-title "Example" --annotate \
  --path /tmp/peekaboo-example.png --json

Read the returned elements before acting. Keep actions tied to the intended app, window, and fresh snapshot; do not reuse example element IDs. After an action, inspect the UI again and check the requested result. Refresh the observation when the window changes or a target becomes stale.

For a screenshot without element detection:

peekaboo see --mode screen --screen-index 0 --no-elements --retina \
  --path /tmp/peekaboo-screen.png --json

Raw keyboard chords require explicit foreground interaction or a fresh exact-window receipt accepted by press. Dragging always moves the shared cursor and requires --foreground. When foreground interaction is authorized:

peekaboo press cmd+shift+t --app Safari --foreground --json
peekaboo drag --from 100,500 --to 100,200 --duration 800ms --foreground --json

Use the actual target and coordinates from the current observation. Prefer explicit duration units such as 800ms; consult each command's help for its options and prerequisites.

Troubleshooting

  • For a removed-command error, use the replacement named by the installed CLI.
  • For host or permission failures, inspect peekaboo bridge status --json and peekaboo permissions status --json for the selected host.
  • For targeting failures, read the leaf help and obtain fresh UI state with see.
  • On macOS 15+, the private window picker prompt is separate from the base Screen Recording grant and can appear even when Bridge permissions are correct.
SKILL.md on GitHub

Sponsors

SnugThe clipboard manager you and your agent can use, safelyAgent CaffeineClose your lid, come back to finished workWaxBlock anything with supervision, without erasing your iPhoneOpen StepsSponsoredFree and open-source skills that make AI coding agents easier to understand, verify, and control.

Details

Repository
github.com/steipete/clawdis

More from steipete/clawdis

All 61
  • Apple Reminders3.2K
  • Clawhub3.2K
  • Coding Agent3.2K
  • Himalaya3.2K
  • Obsidian3.2K
  • 1password3.1K
  • Apple Notes3.1K
  • Video Frames3.1K
  • Healthcheck2.9K
  • Session Logs2.9K
  • Trello2.9K
  • Blogwatcher2.8K
  • Github2.8K
  • Model Usage2.8K
  • Sonoscli2.8K
  • Notion2.7K
  • Openai Whisper Api2.7K
  • Openhue2.7K
  • Sherpa Onnx Tts2.7K
  • Spotify Player2.7K
  • Things Mac2.7K
  • Bear Notes2.6K
  • Blucli2.6K
  • Camsnap2.6K

Recommended

1juliusbrussee avatarcavemanjuliusbrussee/cavemanUltra-compressed communication mode. Cuts output tokens 65% (measured) by speaking like caveman while keeping full technical accuracy. Supports intensity levels594.9Kinstalls
2mattpocock avatargrill-memattpocock/skillsA relentless interview to sharpen a plan or design.1.4Minstalls
3shadcn avatarimproveshadcn/improveSurvey any codebase as a senior advisor and produce prioritized, self-contained implementation plans for OTHER models/agents to execute. Strictly read-only on s37.9Kinstalls
4obra avatarsystematic-debuggingobra/superpowersUse when encountering any bug, test failure, or unexpected behavior, before proposing fixes309.2Kinstalls
5forrestchang avatarkarpathy-guidelinesforrestchang/andrej-karpathy-skillsBehavioral guidelines to reduce common LLM coding mistakes through explicit assumptions, simplicity, and verifiable success criteria.17.9Kinstalls
6vercel-labs avatarfind-skillsvercel-labs/skillsHelps users discover and install agent skills when they ask questions like "how do I do X", "find a skill for X", "is there a skill that can...", or express int4.1Minstalls