de en es ja ko ru zh

MCP Screenshot Server

Name: MCP Screenshot Server
Author: sethbang

by sethbang

Overview Schema Source Related Servers Discussions Admin

TypeScript

Hybrid

Universal Screenshot MCP

npm version MCP Registry License

Ein MCP (Model Context Protocol)-Server, der KI-Assistenten Screenshot-Funktionen bereitstellt – sowohl für die Aufnahme von Webseiten über Puppeteer als auch für plattformübergreifende System-Screenshots mithilfe nativer Betriebssystem-Tools.

Funktionen

Webseiten-Screenshots — Erfassen Sie jede öffentliche URL mit einem Headless-Chromium-Browser
Plattformübergreifende System-Screenshots — Vollbild-, Fenster- oder Bereichsaufnahmen mit nativen OS-Tools (macOS screencapture, Linux maim/scrot/gnome-screenshot/etc., Windows PowerShell+.NET)
Sicherheitsorientiertes Design — SSRF-Prävention, Schutz vor Pfad-Traversal, Schutz vor DNS-Rebinding, Verhinderung von Befehlsinjektionen und DoS-Begrenzung
MCP-nativ — Integriert sich direkt in Claude Desktop, Cursor und jeden MCP-kompatiblen Client

Related MCP server: Chrome DevTools MCP

Anforderungen

Node.js >= 18.0.0
Chromium wird beim ersten Start automatisch von Puppeteer heruntergeladen

Plattformspezifische Anforderungen für `take_system_screenshot`

Plattform	Erforderliche Tools	Hinweise
macOS	`screencapture` (integriert)	Keine zusätzliche Installation erforderlich
Linux	Eines der folgenden: `maim`, `scrot`, `gnome-screenshot`, `spectacle`, `grim` oder `import` (ImageMagick)	`maim` oder `scrot` für volle Funktionsunterstützung empfohlen. Für die Aufnahme von Fenstern nach Namen zusätzlich `xdotool` installieren.
Windows	`powershell` (integriert)	Verwendet .NET `System.Drawing` — keine zusätzliche Installation erforderlich

Linux-Installationsbeispiele

# Ubuntu/Debian (recommended)
sudo apt install maim xdotool

# Fedora
sudo dnf install maim xdotool

# Arch Linux
sudo pacman -S maim xdotool

# Wayland (Sway, etc.)
sudo apt install grim

Nach der Installation können Sie Ihr Setup wie folgt überprüfen:

npx universal-screenshot-mcp --doctor

Dies untersucht den Host und gibt kopierbare Installationsbefehle für fehlende Tools aus, angepasst an Ihre erkannte Distribution.

Schnelleinstieg

Installation über npm

npm install -g universal-screenshot-mcp

Oder führen Sie es direkt mit npx aus:

npx universal-screenshot-mcp

Installation aus dem Quellcode

git clone https://github.com/sethbang/mcp-screenshot-server.git
cd mcp-screenshot-server
npm install
npm run build

Konfigurieren Sie Ihren MCP-Client

Fügen Sie den Server zur Konfiguration Ihres MCP-Clients hinzu. Für Claude Desktop bearbeiten Sie ~/Library/Application Support/Claude/claude_desktop_config.json:

{
  "mcpServers": {
    "screenshot-server": {
      "command": "npx",
      "args": ["-y", "universal-screenshot-mcp"]
    }
  }
}

Oder bei Installation aus dem Quellcode:

{
  "mcpServers": {
    "screenshot-server": {
      "command": "node",
      "args": ["/absolute/path/to/mcp-screenshot-server/build/index.js"]
    }
  }
}

Für Claude Code registrieren Sie den Server mit dem Befehl claude mcp add:

# Project scope (current directory only)
claude mcp add screenshot-server -- npx -y universal-screenshot-mcp

# User scope (available across all projects)
claude mcp add --scope user screenshot-server -- npx -y universal-screenshot-mcp

Oder bei Installation aus dem Quellcode:

claude mcp add screenshot-server -- node /absolute/path/to/mcp-screenshot-server/build/index.js

Überprüfen Sie die Registrierung des Servers mit claude mcp list oder prüfen Sie den Live-Status innerhalb einer Sitzung mit /mcp.

Für Cursor oder andere MCP-Clients konsultieren Sie deren Dokumentation für die entsprechende Konfiguration.

Tools

Der Server stellt zwei MCP-Tools bereit:

`take_screenshot`

Erfasst eine Webseite (oder ein spezifisches Element) über einen Headless-Puppeteer-Browser.

Parameter	Typ	Erforderlich	Beschreibung
`url`	string	✅	Zu erfassende URL (nur http/https)
`width`	number	—	Viewport-Breite (1–3840)
`height`	number	—	Viewport-Höhe (1–2160)
`fullPage`	boolean	—	Erfasst die gesamte scrollbare Seite
`selector`	string	—	CSS-Selektor zur Erfassung eines spezifischen Elements
`waitForSelector`	string	—	Vor der Aufnahme auf diesen Selektor warten
`waitForTimeout`	number	—	Verzögerung in Millisekunden (0–30000)
`outputPath`	string	—	Ausgabedateipfad (Standard: `~/Documents/screenshots`)

Beispiel-Prompt:

Mache einen Screenshot von https://example.com mit 1920x1080

`take_system_screenshot`

Erfasst den Desktop, ein spezifisches Anwendungsfenster oder einen Bildschirmbereich mithilfe nativer OS-Tools. Funktioniert unter macOS, Linux und Windows.

Parameter	Typ	Erforderlich	Beschreibung
`mode`	enum	✅	`fullscreen`, `window` oder `region`
`windowId`	number	—	Fenster-ID für den Fenstermodus
`windowName`	string	—	App-Name (z. B. `"Safari"`, `"Firefox"`) für den Fenstermodus
`region`	object	—	`{ x, y, width, height }` für den Bereichsmodus
`display`	number	—	Display-Nummer für Multi-Monitor-Setups
`includeCursor`	boolean	—	Mauszeiger in die Aufnahme einbeziehen
`format`	enum	—	`png` (Standard) oder `jpg`
`delay`	number	—	Aufnahmeverzögerung in Sekunden (0–10)
`outputPath`	string	—	Ausgabedateipfad (Standard: `~/Documents/screenshots`)

Plattformübergreifende Funktionsunterstützung

Funktion	macOS	Linux	Windows
Vollbild	✅	✅	✅
Bereich	✅	✅ (maim, scrot, grim, import)	✅
Fenster nach Name	✅	⚠️ X11 + xdotool	⚠️ best-effort
Fenster nach ID	✅	✅ nur X11	⚠️ HWND
Multi-Display	✅	⚠️ tool-abhängig	✅
Cursor einbeziehen	✅	⚠️ tool-abhängig	⚠️
Verzögerung	✅	✅	✅

Beispiel-Prompt:

Mache einen System-Screenshot des Safari-Fensters

Konfiguration

Umgebungsvariablen

Variable	Standard	Beschreibung
`SCREENSHOT_OUTPUT_DIR`	`Documents/screenshots`	Standard-Ausgabeverzeichnis relativ zu `~`
`ALLOW_LOCAL`	`false`	Auf `true` setzen, um Screenshots von `localhost`/`127.x.x.x`/`[::1]` zu erlauben (nützlich für lokale Entwicklungsserver)

Ausgabeverzeichnisse

Screenshots werden standardmäßig unter ~/Documents/screenshots gespeichert (konfigurierbar über SCREENSHOT_OUTPUT_DIR). Benutzerdefinierte Ausgabepfade müssen in eines dieser erlaubten Verzeichnisse aufgelöst werden:

Verzeichnis	Beschreibung
`~/Documents/screenshots`	Standard-Ausgabeort (konfigurierbar)
`~/Desktop/Screenshots`	Ursprünglicher Standardort
`~/Downloads`	Benutzer-Download-Ordner
`~/Documents`	Benutzer-Dokumente-Ordner
`/tmp`	System-Temporärverzeichnis

Sicherheit

Dieser Server implementiert mehrere Ebenen der Sicherheitsabsicherung:

ID	Bedrohung	Gegenmaßnahme
SEC-001	SSRF / DNS-Rebinding	URLs werden gegen blockierte IP-Bereiche validiert; DNS wird vor der Anfrage aufgelöst mit IP-Pinning via `--host-resolver-rules`; Navigations-Redirects werden erneut validiert
SEC-003	Befehlsinjektion	Alle Subprozesse verwenden `execFile` (keine Shell); App-Namen werden gegen `SAFE_APP_NAME_PATTERN` validiert
SEC-004	Pfad-Traversal	Ausgabepfade werden mit `fs.realpath()` Symlink-Auflösung validiert; auf erlaubte Verzeichnisse beschränkt
SEC-005	Denial of Service	Gleichzeitige Puppeteer-Instanzen auf 3 via Semaphor begrenzt

Für vollständige Details siehe docs/security.md.

Entwicklung

Skripte

Befehl	Beschreibung
`npm run build`	Kompiliert TypeScript nach `build/`
`npm run watch`	Rekompiliert bei Dateiänderungen
`npm test`	Unit-Tests (schnell, vollständig gemockt)
`npm run test:integration`	Integrationstests (echtes DNS/Dateisystem)
`npm run test:e2e`	E2E-Tests (echtes Puppeteer/native Tools)
`npm run test:all`	Alle Testebenen zusammen
`npm run test:linux`	Linux E2E via Docker (erfordert Docker)
`npm run test:watch`	Tests im Watch-Modus ausführen
`npm run test:coverage`	Tests mit Coverage-Bericht ausführen
`npm run lint`	Quellcode mit ESLint prüfen
`npm run inspector`	Startet MCP Inspector zum Debuggen

Projektstruktur

src/
├── index.ts                 # Entry point — stdio transport
├── server.ts                # MCP server factory
├── config/
│   ├── index.ts             # Static constants (limits, allowed dirs)
│   └── runtime.ts           # Singleton semaphore, default directory
├── tools/
│   ├── take-screenshot.ts   # Web page capture tool
│   └── take-system-screenshot.ts  # macOS system capture tool
├── types/
│   └── index.ts             # Shared TypeScript interfaces
├── utils/
│   ├── helpers.ts           # Response builders, file utilities
│   ├── screenshot-provider.ts # Cross-platform provider interface + factory
│   ├── macos-provider.ts    # macOS: screencapture wrapper
│   ├── linux-provider.ts    # Linux: maim/scrot/gnome-screenshot/etc.
│   ├── windows-provider.ts  # Windows: PowerShell + .NET System.Drawing
│   ├── macos.ts             # Window ID lookup via CoreGraphics
│   └── semaphore.ts         # Async concurrency limiter
└── validators/
    ├── path.ts              # Output path validation (SEC-004)
    └── url.ts               # URL/SSRF validation (SEC-001)

Testen

Tests verwenden Vitest in drei Ebenen:

Unit (npm test) — Vollständige Dependency Injection, kein echtes I/O. Schnelle Feedbackschleife.
Integration (npm run test:integration) — Echte DNS-Auflösung, echtes Dateisystem mit temporären Verzeichnissen, echtes Puppeteer gegen einen lokalen HTTP-Server.
E2E (npm run test:e2e) — Echte native Screenshot-Tools. macOS-Tests laufen nativ; Linux-Tests laufen in Docker via npm run test:linux.

npm test                 # Unit tests (~300ms)
npm run test:linux       # Linux provider tests in Docker
npm run test:all         # Everything

Debuggen mit dem MCP Inspector

npm run inspector

Dies startet den MCP Inspector, der mit Ihrem gebauten Server verbunden ist, sodass Sie Tools interaktiv aufrufen können.

Lizenz

Available Tools

2 tools

take_screenshotB

Capture web page or element via headless browser. Saves to ~/Documents/screenshots by default (configurable via SCREENSHOT_OUTPUT_DIR env var).

ParametersJSON Schema

Name	Required	Description
`url`	Yes	URL to capture
`width`	No	Viewport width
`height`	No	Viewport height
`fullPage`	No	Capture full page
`selector`	No	CSS selector for element
`waitForSelector`	No	Wait for selector
`waitForTimeout`	No	Delay in ms
`outputPath`	No	Absolute path, or relative to home dir

TDQS

B3.4/5.0

Behavior2/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

With no annotations, the description carries full burden but only mentions method and save location; it omits critical behavioral traits like destructiveness, permission needs, or return value.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness5/5

Is the description appropriately sized, front-loaded, and free of redundancy?

Two concise sentences: first states purpose, second details save behavior. Front-loaded and no wasted words.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness2/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

Despite 8 parameters and no output schema, the description omits return value, error handling, and other essential context, leaving significant gaps.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters3/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema covers all 8 parameters with descriptions, so baseline is 3. The description adds the default save path and env var override, providing marginal extra context.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

The description uses specific verb 'capture' and resource 'web page or element' with method 'via headless browser', clearly distinguishing it from sibling 'take_system_screenshot' which captures system screens.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines3/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

The description implies usage for web screenshots but does not explicitly state when to use versus alternatives, nor does it mention prerequisites or exclusions.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

take_system_screenshotA

Capture desktop, window, or region screenshot. Cross-platform: macOS (screencapture), Linux (maim/scrot/gnome-screenshot/etc.), Windows (PowerShell+.NET). Saves to ~/Documents/screenshots by default (configurable via SCREENSHOT_OUTPUT_DIR env var). For window mode, provide windowName (app name like "Safari") or windowId.

ParametersJSON Schema

Name	Required	Description
`mode`	Yes	fullscreen=entire screen, window=specific app (requires windowName or windowId), region=coordinates
`windowId`	No	Window ID (for window mode)
`windowName`	No	App name like "Safari", "Firefox" (for window mode)
`region`	No	Region {x,y,width,height}
`display`	No	Display number
`includeCursor`	No	Include cursor
`format`	No	Image format (png or jpg)
`delay`	No	Delay seconds
`outputPath`	No	Absolute path, or relative to home dir

TDQS

A3.6/5.0

Behavior3/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

No annotations are provided, so the description must carry the full burden. It discloses the default save location and cross-platform tool dependencies but does not mention return behavior (e.g., file path or binary), permission requirements, or failure modes.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness4/5

Is the description appropriately sized, front-loaded, and free of redundancy?

The description is concise with 4 sentences, front-loading the main action. It effectively uses bullet-like information for cross-platform details and mode instructions, though it could be slightly more structured.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness3/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

Given 9 parameters including a nested object and no output schema, the description covers default directory and platform support but omits return value, error handling, and comparison with the sibling tool. It is adequate but not fully complete.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters3/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema description coverage is 100%, so the baseline is 3. The description adds minor value by specifying default output directory and that windowName is an app name, but the schema already describes all parameters adequately.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

The description clearly states 'Capture desktop, window, or region screenshot' with specific verb and resource. It distinguishes from the sibling tool 'take_screenshot' by specifying cross-platform support and modes.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines3/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

The description provides clear context for when to use each mode (fullscreen, window, region) and gives examples for window mode. However, it lacks guidance on when not to use this tool versus the sibling 'take_screenshot', missing explicit exclusion or alternative differentiation.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

2 tool updatesv1.2.0May 6, 2026

Changedtake_screenshot10 fields changed

addedInput schema / $schema

Added value: +"http://json-schema.org/draft-07/schema#"

addedInput schema / additionalProperties
```
Added value: +false
```

changedInput schema / properties / fullPage / description

Previous value: -"Capture full scrollable page"New value: +"Capture full page"

changedInput schema / properties / height / description

Previous value: -"Viewport height in pixels"New value: +"Viewport height"

changedInput schema / properties / outputPath / description

Previous value: -"Custom output path (optional)"New value: +"Absolute path, or relative to home dir"

addedInput schema / properties / selector

Added value: +{
+  "description": "CSS selector for element",
+  "type": "string"
+}

changedInput schema / properties / url / description

Previous value: -"URL to capture (can be http://, https://, or file:///)"New value: +"URL to capture"

addedInput schema / properties / waitForSelector

Added value: +{
+  "description": "Wait for selector",
+  "type": "string"
+}

addedInput schema / properties / waitForTimeout

Added value: +{
+  "description": "Delay in ms",
+  "maximum": 30000,
+  "minimum": 0,
+  "type": "number"
+}

changedInput schema / properties / width / description

Previous value: -"Viewport width in pixels"New value: +"Viewport width"

Addedtake_system_screenshot

1 tool updateAug 18, 2025
- First observedtake_screenshot

Deploy Server

Give agents eyes on the webHosted screenshot MCP. Send a URL, get a pixel-perfect image. $0.002 per grab. Works with Claude, Cursor, and Codex.grabbit.liveAd

TDQS

A3.7/5.0

Scored 2026-05-25 20:15 across 2 tools

Disambiguation5/5

The two tools are clearly distinct: one captures web pages/elements via headless browser, the other captures desktop/system screenshots. There is no overlap in functionality.

Naming Consistency5/5

Both tools follow a consistent verb_noun pattern with snake_case (take_screenshot, take_system_screenshot), using the same 'take_' prefix.

Tool Count3/5

Only 2 tools for a screenshot server is minimal but still covers the core use cases. More tools might be expected for management or configuration, but the count is not severely inadequate.

Completeness4/5

The tool set covers the primary screenshot domains (web and system). Minor gaps exist, such as listing or deleting screenshots, but core functionality is well-covered.

Maintenance

ActivityMaintained

ResponsivenessNo issues

Resources

Need Help?

Reddit Discussion

Related Servers

Related MCP Connectors

Browserless A
io.github.pipeworx-io
Browserless MCP — wraps the Browserless headless-Chromium REST API (browserless.io)
Browser Automation Web Scraping
401
screenshot A
com.apiscreenshot
Screenshot any URL/HTML as PNG/JPEG/WebP, or read it as clean Markdown/text for LLMs.
Browser Automation Image & Video Processing
2
Framejet Screenshot A
dev.framejet
Clean PNG/JPEG screenshots via REST or MCP, with goal-driven multi-step navigation.
Browser Automation Web Scraping
11
Skillforge Render API A
com.skillforge99
Screenshot, PDF, OG-image, and page extraction (markdown/JSON) over MCP. Bearer key or x402.
Image & Video Processing Multimedia Processing Marketing
41

Related MCP Servers

Webpage Screenshot MCP Server
Web Scraping Browser Automation
ananddtyagi
A
license
A
quality
D
maintenance
Captures screenshots of web pages using Puppeteer, allowing AI agents to visually verify web applications and see their progress when generating web apps.
Updated a year ago (2025-06-29 03:57 UTC)
5
57
MIT
Chrome DevTools MCPofficial
Browser Automation Web Scraping Testing & QA Tools
ChromeDevTools
A
license
Not graded
quality
A
maintenance
Enables AI coding assistants to control and inspect a live Chrome browser through Chrome DevTools. Provides browser automation, performance analysis, debugging capabilities, and network request monitoring.
Updated a day ago (2026-10-01 15:10 UTC)
2,487,841 npm
52,860
Apache 2.0
Webshot MCP
Browser Automation Web Scraping Multimedia Processing
bingal
A
license
C
quality
D
maintenance
Enables taking screenshots of web pages with support for multiple devices (desktop, mobile, tablet), custom dimensions, full-page capture, and various image formats. Built with Playwright for reliable web page rendering and screenshot generation.
Updated a year ago (2025-09-30 04:08 UTC)
1
MIT
Screenshot API
Browser Automation Web Scraping
raphy78626
A
license
Not graded
quality
B
maintenance
Enables AI agents to capture any public URL as PNG, JPEG, or PDF via REST API or MCP tools, including screenshot capture, page description, and PDF rendering.
Updated 2 months ago (2026-08-07 10:50 UTC)
16 npm
MIT

Universal Screenshot MCP

Funktionen

Anforderungen

Plattformspezifische Anforderungen für take_system_screenshot

Linux-Installationsbeispiele

Schnelleinstieg

Installation über npm

Installation aus dem Quellcode

Konfigurieren Sie Ihren MCP-Client

Tools

take_screenshot

take_system_screenshot

Plattformübergreifende Funktionsunterstützung

Konfiguration

Umgebungsvariablen

Ausgabeverzeichnisse

Sicherheit

Entwicklung

Skripte

Projektstruktur

Testen

Debuggen mit dem MCP Inspector

Lizenz

Available Tools

TDQS

TDQS

Tool Schema Changelog

TDQS

Maintenance

Resources

Related MCP Connectors

Related MCP Servers

Webpage Screenshot MCP Server

Chrome DevTools MCPofficial

Webshot MCP

Screenshot API

Appeared in Searches

Plattformspezifische Anforderungen für `take_system_screenshot`

`take_screenshot`

`take_system_screenshot`