Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add skills/ruchit-p/macwright/macwright-patternsnpx skills add ruchit-p/macwright --skill macwright-patternsgit clone --depth 1 https://github.com/ruchit-p/macwrightWhat it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00093 | $0.02879 |
| Opus 5 | $0.00046 | $0.01439 |
| Sonnet 5 | $0.00019 | $0.00576 |
| Haiku 4.5 | $0.00009 | $0.00288 |
Grade A, and why
macwright-patterns scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured yesterday.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 236 lines — stays where its author put it; the contents beside it link to each section on GitHub.
MacWright Automation Patterns
Patterns and pitfalls discovered from iterative testing of all 24 macwright tools.
Wait+Screenshot Chaining
Most action tools accept optional wait (ms) and screenshot (bool) params:
click(x: 500, y: 300, wait: 500, screenshot: true)
→ clicks, waits 500ms, returns screenshot
This collapses 3 tool calls into 1. Use it to verify actions immediately.
Tool Selection Guide
| Task | Best Tool | Notes |
|---|---|---|
| Navigate Safari | safari_url |
Never click address bar |
| Manage Safari tabs | safari_tabs |
list/new/close/select actions |
| Fill web form | fill_form |
Multi-field, dispatches input/change events, supports textarea |
| Click web element | click_element |
CSS selector → scrolls into view → click in one step |
| Click by visible text | click_text("Submit") |
No CSS selector — finds by text, picks most specific match |
| Hover over element | hover_element |
CSS selector → moves mouse over it (triggers :hover, dropdowns) |
| Get web element coords | find_element |
When you need coords without clicking |
| Find all matching elements | find_element(all: true) |
Returns array of coords + text |
| Screenshot a DOM element | screenshot_element |
Scrolls into view, captures region |
| Wait for DOM element | wait_for_element |
After navigation/AJAX |
| Scroll web page (relative) | scroll_page(y: 500) |
Focus-free, returns position |
| Scroll web page (absolute) | scroll_page(y: 0, absolute: true) |
Jump to exact position (e.g. top) |
| Scroll overflow div element | scroll_element(selector, y: 200) |
Scroll a specific DOM container |
| Scroll native app | scroll + focus_window first |
App must be frontmost |
| Type ASCII | type_text |
Direct AppleScript |
| Type emoji/Unicode | type_text |
Auto-detects, uses clipboard |
| Type with autocomplete | type_text(slowly: true) |
50ms delay per char, triggers keystroke handlers |
| Run shell command | run_shell |
Returns stdout/stderr; exit code in text, only timeout = isError |
| Process data with jq | run_shell(command: "jq ...") |
jq available at /usr/bin/jq |
| Process data with python | run_shell(command: "python3 -c ...") |
[email protected] available |
| Reload Safari page | safari_reload |
Faster than Cmd+R |
| Check what's visible | screenshot(x,y,width,height) |
Region param for focused view |
| Find app state | get_frontmost_app |
Returns bounds too |
| List all windows | list_windows |
JSON array of apps+bounds |
| Dismiss menus | press_key(key: "esc") |
Works for context menus too |
| Window positioning | resize_window |
Set exact x,y,width,height |
| Current page info | get_page_info() |
URL, title, readyState in one call |
| Discover form fields | get_form_fields() |
All inputs/selects/buttons with types/names |
| Extract all links | get_links(filter: '/blog') |
Links with text+href, filterable by URL |
| Scroll to element | scroll_to_element(selector) |
scrollIntoView with alignment |
| Focus element | focus_element(selector) |
Trigger focus event (no click) |
| Navigate forward | safari_navigate_forward |
History.forward() |
| Wait for condition | wait_for_function(expr) |
Most flexible wait — any JS expression |
| Wait for text | wait_for_text("Success") |
Page body textContent check |
| Wait for URL | wait_for_url("/dashboard") |
URL contains pattern |
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- yesterday First seen · 236 lines · 93 tokens per session scan A 63ecd91b7c17
macwright-patterns is a skill published in the GitHub repository ruchit-p/macwright (0 stars, last pushed 20d ago), licensed MIT. It adds 93 tokens to every session and 2,879 once invoked, about $0.0005 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
macos-harness
Control a whole Mac from one persistent Python session with screenshots, PID-targeted input, an animated virtual pointer, targeted Apple Accessibility, Apple Events, Browser Harness CDP, and filesystem access. Use for native, Electron, browser, dialog, file, or cross-app tasks without moving the physical cursor or…
web-navigator
Web 浏览器自动化路由规则。通过 webnavigator 将 Web 场景路由到 Playwright MCP 或 mcp-chrome。.
peekaboo
Use Peekaboo for macOS desktop automation, screenshots, visual UI maps, native accessibility inspection, app/window/menu/dialog control, native app and browser chrome control, browser-page MCP tooling, MCP diagnostics, and Peekaboo repo validation. Use when Codex needs current macOS UI state, direct desktop control…
performance
Optimize web performance for faster loading and better user experience. Use when asked to "speed up my site", "optimize performance", "reduce load time", "fix slow loading", "improve page speed", or "performance audit".
core-web-vitals
Optimize Core Web Vitals (LCP, INP, CLS) for better page experience using field and lab evidence. Use when asked to "improve Core Web Vitals", "fix LCP", "reduce CLS", "optimize INP", "page experience optimization", or "fix layout shifts".
scarf-miniapp-author
Author a Scarf mini-app — a small sandboxed web surface (HTML/CSS/JS) that renders inside a project's cockpit and talks to the bound Hermes session + project data through the versioned window.scarf bridge. Use to build a bespoke panel (a task board, an approval queue, a chart, a data table) for a project.