safari-devtools-mcp
Controls and inspects a live Safari browser on macOS for debugging, automation, and testing, including console log capture, network request interception, DOM inspection, and more.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@safari-devtools-mcp Show me the console logs and network requests on the current Safari page."
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Safari DevTools MCP
safari-devtools-mcp lets your coding agent (such as Claude, Cursor, Copilot or Gemini) control and inspect a live Safari browser on macOS. It acts as a Model-Context-Protocol (MCP) server, giving your AI coding assistant access to Safari DevTools for debugging, automation, and testing.
Chrome developers get powerful AI debugging through chrome-devtools-mcp by Google. Safari developers should have something equivalent — this project brings that same experience to Safari, with matching tool names and parameter schemas so you can swap between browsers with minimal friction.
Changelog | Contributing
Note: This server exposes browser content (page data, console logs, network traffic) to MCP clients. Avoid browsing sensitive websites or entering credentials while a session is active.
Related MCP server: Safari MCP Server
Why safari-devtools-mcp?
This project uses WebDriver for capabilities that scripting alone cannot provide:
Network request/response body capture — intercepts fetch and XHR calls with full headers, payloads, and timing
DOM snapshots via accessibility tree — stable element UIDs that survive page re-renders, not brittle CSS selectors
CSS computed style inspection — read any computed property from any element
Cookie and storage management — read, write, and delete cookies, localStorage, and sessionStorage
Element-level screenshots — capture individual elements, not just the full viewport
Session auto-recovery — detects dead SafariDriver sessions and reconnects transparently
Key features
Browser debugging: Capture console logs, inspect network requests, and evaluate JavaScript — with stack traces and full request/response details.
Reliable automation: Click, type, fill forms, drag and drop, and press keyboard shortcuts using accessibility-tree snapshots with stable UIDs.
Native macOS integration: Tab management via AppleScript for listing, switching, and controlling Safari tabs across windows
Requirements
macOS (Safari and SafariDriver are Apple-exclusive)
Node.js 18+ (22+ recommended)
Safari with remote automation enabled
Safari setup
Open Safari > Settings > Advanced > check "Show features for web developers"
Develop menu > check "Allow Remote Automation"
Authorize SafariDriver:
sudo safaridriver --enable
Getting started
Standard MCP config:
{
"mcpServers": {
"safari-devtools": {
"command": "npx",
"args": ["-y", "safari-devtools-mcp@latest"]
}
}
}claude mcp add safari-devtools -- npx -y safari-devtools-mcp@latestOr add to your project's .mcp.json using the standard config above.
Add the standard config to ~/Library/Application Support/Claude/claude_desktop_config.json.
Add the standard config to your Cursor MCP settings.
Add the standard config to .vscode/mcp.json.
copilot mcp add safari-devtools -- npx -y safari-devtools-mcp@latestgemini mcp add safari-devtools -- npx -y safari-devtools-mcp@latestOr add the standard config to your ~/.gemini/settings.json.
Add the standard config to your .gemini/settings.json in the project root.
Add the standard config to your .junie/mcp.json in the project root.
Open "Install MCP Server" in Raycast and fill in:
Command:
npxArguments:
-y safari-devtools-mcp@latest
Or copy the standard config JSON above before opening the command — Raycast will auto-fill the form.
Add the standard config to your .vs/mcp.json in the solution root.
Add the standard config to your Warp MCP settings file at ~/.warp/mcp.json.
git clone https://github.com/HayoDev/safari-devtools-mcp.git
cd safari-devtools-mcp
npm install && npm run buildThen point your MCP client to the built entry point:
{
"mcpServers": {
"safari-devtools": {
"command": "node",
"args": [
"/path/to/safari-devtools-mcp/build/src/bin/safari-devtools-mcp.js"
]
}
}
}Your first prompt
Navigate to https://example.com, take a snapshot, and list any console errors.
Slim mode
Use --slim to reduce token usage with shorter tool descriptions. Useful when working with context-limited models or long conversations:
{
"mcpServers": {
"safari-devtools": {
"command": "npx",
"args": ["-y", "safari-devtools-mcp@latest", "--slim"]
}
}
}Prompts (skills)
The server exposes guided debugging workflows as MCP prompts. Clients that support prompts (e.g. Claude Desktop, Claude Code) can invoke them by name:
Prompt | Description |
| Getting started — tool overview, recommended workflow, and Safari-specific notes |
| Accessibility audit — a11y tree inspection, axe-core injection, WCAG checks |
| Debug WebKit quirks — CSS prefixes, JS feature gaps, ITP/CORS issues |
| Performance analysis — Navigation Timing, Core Web Vitals, resource waterfall |
Tools (45)
Debugging
Tool | Description |
| List console messages with filtering by level (log, warn, error) |
| Get a detailed message including stack trace and arguments |
| Clear all captured console messages |
| Monitor network requests — Fetch, XHR, and resource loads |
| Get full request/response details with headers and body |
| Clear all captured network requests |
| Execute JavaScript in the browser context and return results |
| Capture a PNG screenshot of the page or a specific element |
| Accessibility-tree snapshot of the DOM with stable UIDs for element targeting |
Page content
Tool | Description |
| Get the page title, URL, and visible text content |
| Get the full HTML source of the page |
| Extract all links with their text, href, and rel attributes |
| Extract meta tags (og:, twitter:, description, etc.) |
Navigation
Tool | Description |
| List all open Safari tabs across windows |
| Switch to a specific tab |
| Open a new tab and navigate to a URL |
| Close a tab |
| Navigate to a URL, go back, forward, or reload |
| Wait for specific text to appear on the page |
| Resize the browser window |
| Accept or dismiss browser dialogs (alert, confirm, prompt) |
Scroll
Tool | Description |
| Scroll the page in any direction by a given amount |
| Scroll an element into view by its UID |
CSS inspection
Tool | Description |
| Get computed CSS styles for any element by UID |
Cookies & storage
Tool | Description |
| Get browser cookies, optionally filtered by name or domain |
| Set a cookie with name, value, and optional attributes |
| Delete a cookie by name, or delete all cookies |
| Read from localStorage or sessionStorage |
| Write a key-value pair to localStorage or sessionStorage |
| Delete a key or clear all entries from localStorage/sessionStorage |
Input automation
Tool | Description |
| Click an element by UID from a snapshot |
| Click at specific x/y coordinates |
| Right-click (context menu) on an element |
| Select an option from a dropdown by value or label |
| Hover over an element |
| Type into an input field or select from a dropdown |
| Fill multiple form fields at once |
| Type text into the currently focused element |
| Drag and drop between elements or coordinates |
| Press a key or combination (e.g., |
| Upload a file through a file input |
iOS Safari validation
Tool | Description |
| Parse the viewport meta tag and validate against iOS best practices (width, zoom, viewport-fit) |
| Read CSS safe-area-inset values and check whether the page handles notched devices correctly |
| Audit the page for Add to Home Screen / PWA readiness (apple-touch-icon, manifest, splash screens, status bar) |
| Check page CSS against the live Safari session via CSS.supports() |
Architecture
MCP Client (Claude, Cursor, etc.)
| stdio (MCP protocol)
v
Safari DevTools MCP Server
|
v
+-------------------------------+
| SafariDriver |
| +-- Selenium WebDriver | <-- Browser automation
| +-- JS Injection | <-- Console/Network capture
| +-- AppleScript (osascript) | <-- Native tab management
+-------------------------------+
|
v
Safari BrowserKnown limitations
Single session: Safari only allows one WebDriver session at a time. Running multiple instances is not supported.
macOS only: Safari and SafariDriver are Apple-exclusive — this server does not work on Linux or Windows.
No headless mode: Safari does not support headless operation. A visible browser window is required.
Console pre-injection gap: Console messages logged before the capture script is injected are not captured.
Network pre-injection gap: Network requests made before injection are backfilled via the Performance API with limited detail (timing and size, but no headers or status codes).
PNG only: Safari WebDriver only supports PNG screenshots — no JPEG or WebP. Full-page capture is not available.
AppleScript permissions: Tab management features require macOS Accessibility permissions for
osascript.
Credits
chrome-devtools-mcp by Google/ChromeDevTools — the inspiration and interface standard this project mirrors. Tool names and schemas are intentionally compatible so you can switch between Chrome and Safari seamlessly.
safari-mcp-server — prior art for Safari MCP integration that we studied during development.
License
Available Tools
49 toolscheck_ios_web_app_readinessA
Audit the page for iOS Safari "Add to Home Screen" / PWA readiness. Checks apple-touch-icon, apple-mobile-web-app-capable, status bar style, theme-color, manifest link, and splash screen configuration.
Inputs: no named parameters.
check_webkit_compatibilityA
Check CSS on the current page against the live Safari session. Extracts every CSS property via structured DOM APIs, runs CSS.supports() in the actual browser, and reports what is broken right now — unsupported properties, missing -webkit- prefixes, and known Safari rendering quirks.
Inputs: no named parameters.
clear_consoleA
Clear all captured console messages.
Inputs: no named parameters.
clear_networkA
Clear all captured network requests.
Inputs: no named parameters.
clickC
Clicks on the provided element.
Inputs: uid (required), dblClick.
click_atC
Clicks at the provided coordinates.
Inputs: x (required), y (required), dblClick.
close_pageA
Closes a page by its ID. The last open page cannot be closed.
Inputs: pageId (required).
delete_cookieA
Delete a cookie by name, or delete all cookies when no name is provided.
Inputs: name.
delete_storageA
Delete a key from localStorage or sessionStorage, or clear all entries.
Inputs: key, storageType.
dragB
Drag an element onto another element.
Inputs: to_uid (required), from_uid (required).
evaluate_scriptA
Evaluate a JavaScript function inside the currently selected page. Returns the response as JSON, so returned values have to be JSON-serializable.
Inputs: function (required), args.
extract_linksA
Extract all links from the current page with their text and href.
Inputs: no named parameters.
extract_metaA
Extract meta tags from the current page (og:, twitter:, description, etc.).
Inputs: no named parameters.
fillC
Type text into an input, text area or select an option from a element.
Inputs: uid (required), value (required).
fill_formB
Fill out multiple form elements at once.
Inputs: elements (required).
get_color_schemeA
Check the currently active color scheme preference (light or dark) as seen by the page, and whether an MCP override is active.
Inputs: no named parameters.
get_computed_styleA
Get computed CSS styles for an element by its UID from a snapshot. Returns commonly useful properties by default, or specify exact properties.
Inputs: uid (required), properties.
get_console_messageA
Gets a console message by its ID. You can get all messages by calling list_console_messages.
Inputs: msgid (required).
get_cookiesA
Get browser cookies for the current page. Optionally filter by name or domain.
Inputs: name, domain.
get_html_sourceA
Get the full HTML source of the current page.
Inputs: filePath.
get_network_requestB
Gets a network request by its reqid from the listed requests.
Inputs: reqid (required).
get_page_contentA
Get the page title, URL, and visible text content.
Inputs: no named parameters.
get_safe_area_insetsA
Read the current CSS safe-area-inset values (top, right, bottom, left) as seen by the page. These are non-zero on notched iPhones when viewport-fit=cover is set. Also checks whether the page uses env(safe-area-inset-*) in its stylesheets.
Inputs: no named parameters.
get_storageA
Read from localStorage or sessionStorage. Returns all entries or a specific key.
Inputs: key, storageType.
handle_dialogB
If a browser dialog was opened, use this command to handle it.
Inputs: action (required), promptText.
hoverC
Hover over the provided element.
Inputs: uid (required).
inspect_viewport_metaA
Parse the viewport meta tag and validate it against iOS Safari best practices. Reports issues like missing width=device-width, disabled zoom (accessibility violation), missing viewport-fit=cover for safe areas, and other common misconfiguration.
Inputs: no named parameters.
list_console_messagesA
List all console messages for the currently selected page since the last navigation.
Inputs: types, pageIdx, pageSize.
list_network_requestsA
List all network requests for the currently selected page since the last navigation. Includes historical requests made before monitoring started (with limited detail).
Inputs: pageIdx, pageSize, resourceTypes.
list_pagesC
Get a list of pages (tabs) open in Safari.
Inputs: no named parameters.
new_pageA
Open a new tab and load a URL.
Inputs: url (required).
press_keyA
Press a key or key combination. Use this when other input methods like fill() cannot be used.
Inputs: key (required).
reset_device_emulationA
Remove all device emulation overrides and restore the original viewport size, JS-level user agent, device pixel ratio, and touch settings.
Inputs: no named parameters.
resize_pageB
Resizes the selected page's window so that the page has specified dimensions.
Inputs: width (required), height (required).
right_clickB
Right-click (context click) on the provided element.
Inputs: uid (required).
scrollA
Scroll the page in a direction by a given amount of pixels.
Inputs: direction (required), amount.
scroll_to_elementB
Scroll an element into view by its UID from a snapshot.
Inputs: uid (required).
select_optionA
Select an option from a dropdown by value or visible label.
Inputs: uid (required), label, value.
select_pageA
Select a page as context for future tool calls.
Inputs: pageId (required).
set_color_schemeA
Emulate prefers-color-scheme (light/dark mode) in the current page. Overrides CSS @media rules, the color-scheme property, and JavaScript matchMedia() (including change events) so the page behaves as if the OS preference changed. Use "reset" to restore the system default.
Inputs: colorScheme (required).
set_cookieB
Set a browser cookie with the given name, value, and optional attributes.
Inputs: name (required), value (required), path, domain, +3 more.
set_device_emulationA
Emulate an Apple device by resizing the viewport and overriding device pixel ratio and touch support at the JS level. The user agent is overridden via navigator.userAgent (JS-level only — HTTP request headers are not affected). Choose a preset (iPhone SE, iPhone 14, iPhone 15 Pro, iPhone 16 Pro Max, iPad Mini, iPad Air, iPad Pro 11, iPad Pro 13) or provide custom values. Use reset_device_emulation to restore defaults.
Inputs: width, device, height, userAgent, +1 more.
set_storageA
Write a key-value pair to localStorage or sessionStorage.
Inputs: key (required), value (required), storageType.
take_screenshotB
Take a screenshot of the page or element.
Inputs: uid, filePath.
take_snapshotA
Take a text snapshot of the currently selected page based on the DOM/a11y tree. The snapshot lists page elements along with a unique identifier (uid). Always use the latest snapshot. Prefer taking a snapshot over taking a screenshot.
Inputs: verbose, filePath.
type_textB
Type text using keyboard into a previously focused input.
Inputs: text (required), submitKey.
upload_fileA
Upload a file through a provided file input element.
Inputs: uid (required), filePath (required).
wait_forA
Wait for a condition on the page: text content, a CSS selector, or a JS predicate. Returns a snapshot once the condition is met. At least one condition (text, selector, or predicate) must be provided.
Inputs: text, timeout, visible, selector, +1 more.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 49 tools
Most tools have distinct purposes (e.g., take_screenshot vs take_snapshot), but some overlap exists among input methods like type_text, fill, fill_form, and press_key, which could confuse an agent. Descriptions help differentiate them.
All tools follow a consistent snake_case verb_noun pattern (e.g., list_console_messages, get_network_request, set_cookie). Even compound names like scroll_to_element maintain the pattern. No mixing of conventions.
With 49 tools, the set is far beyond the typical well-scoped range (3-15). While the server attempts to cover many aspects of Safari DevTools, the sheer number makes the surface heavy for an agent to navigate and select effectively.
The tool set covers a wide range of debugging and automation tasks: page lifecycle, console/network inspection, element interaction, storage, cookies, device emulation, accessibility, PWA auditing, and WebKit compatibility. No obvious gaps for the intended domain.
Maintenance
Related MCP Connectors
Control a real browser from AI agents: navigate, click, type, screenshot, read the page.
Hosted browser for agents: open pages, search the web, extract data, read PDFs and take screenshots.
Launch, inspect, control, and share isolated cloud browsers for your agents.
Give AI agents a real cloud Chrome: open sessions, navigate, read pages, click, type and screenshot.
Related MCP Servers
- AlicenseBqualityDmaintenanceProvides AI assistants with Safari browser automation and developer tools access, enabling LLMs to control Safari, access console logs, monitor network activity, and perform browser automation tasks.139 npm33MIT
- AlicenseAqualityCmaintenanceEnables visual web access and browser automation through Safari, with tools for navigation, screenshots, element interaction, and tab management.1719 npmBSD 3-Clause
- AlicenseAqualityDmaintenanceEnables programmatic control of Safari browser on macOS for web automation, testing, and debugging via the Model Context Protocol.167 npm4MIT
- AlicenseNot gradedqualityCmaintenanceEnables coding agents to control and inspect a live Chrome browser for reliable automation, in-depth debugging, and performance analysis via Chrome DevTools.Apache 2.0