Skip to content

Commit 852709c

Browse files
pavelfeldmanpavelfeldman
committed
chore: initial code commit
1 parent b1d5410 commit 852709c

23 files changed

+6307
-204
lines changed

‎.gitignore

+2
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,2 @@
1+
lib/
2+
node_modules/

‎.npmignore

+5
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,5 @@
1+
**/*
2+
README.md
3+
LICENSE
4+
!lib/**/*.js
5+
!cli.js

‎LICENSE

+202-201
Large diffs are not rendered by default.

‎README.md

+223-3
Original file line numberDiff line numberDiff line change
@@ -1,3 +1,223 @@
1-
# Repository setup required :wave:
2-
3-
Please visit the website URL :point_right: for this repository to complete the setup of this repository and configure access controls.
1+
## Playwright MCP
2+
3+
This package is experimental and not yet ready for production use.
4+
It is a subject to change and will not respect semver versioning.
5+
6+
### Example config
7+
8+
```js
9+
{
10+
"mcpServers": {
11+
"playwright": {
12+
"command": "npx",
13+
"args": [
14+
"@playwright/mcp",
15+
"--headless"
16+
]
17+
}
18+
}
19+
}
20+
```
21+
22+
### Running headed browser (Browser with GUI).
23+
24+
```js
25+
{
26+
"mcpServers": {
27+
"playwright": {
28+
"command": "npx",
29+
"args": [
30+
"@playwright/mcp"
31+
]
32+
}
33+
}
34+
}
35+
```
36+
37+
### Running headed browser on Linux
38+
39+
When running headed browser on system w/o display or from worker processes of the IDEs,
40+
you can run Playwright in a client-server manner. You'll run the Playwright server
41+
from environment with the DISPLAY
42+
43+
```sh
44+
npx playwright run-server
45+
```
46+
47+
And then in MCP config, add following to the `env`:
48+
49+
```js
50+
{
51+
"mcpServers": {
52+
"playwright": {
53+
"command": "npx",
54+
"args": [
55+
"@playwright/mcp"
56+
],
57+
"env": {
58+
// Use the endpoint from the output of the server above.
59+
"PLAYWRIGHT_WS_ENDPOINT": "ws://localhost:<port>/"
60+
}
61+
}
62+
}
63+
}
64+
```
65+
66+
### Tool Modes
67+
68+
The tools are available in two modes:
69+
70+
1. **Snapshot Mode** (default): Uses accessibility snapshots for better performance and reliability
71+
2. **Vision Mode**: Uses screenshots for visual-based interactions
72+
73+
To use Vision Mode, add the `--vision` flag when starting the server:
74+
75+
```js
76+
{
77+
"mcpServers": {
78+
"playwright": {
79+
"command": "npx",
80+
"args": [
81+
"@playwright/mcp",
82+
"--vision"
83+
]
84+
}
85+
}
86+
}
87+
```
88+
89+
Vision Mode works best with the computer use models that are able to interact with elements using
90+
X Y coordinate space, based on the provided screenshot.
91+
92+
### Snapshot Mode
93+
94+
The Playwright MCP provides a set of tools for browser automation. Here are all available tools:
95+
96+
- **browser_navigate**
97+
- Description: Navigate to a URL
98+
- Parameters:
99+
- `url` (string): The URL to navigate to
100+
101+
- **browser_go_back**
102+
- Description: Go back to the previous page
103+
- Parameters: None
104+
105+
- **browser_go_forward**
106+
- Description: Go forward to the next page
107+
- Parameters: None
108+
109+
- **browser_click**
110+
- Description: Perform click on a web page
111+
- Parameters:
112+
- `element` (string): Human-readable element description used to obtain permission to interact with the element
113+
- `ref` (string): Exact target element reference from the page snapshot
114+
115+
- **browser_hover**
116+
- Description: Hover over element on page
117+
- Parameters:
118+
- `element` (string): Human-readable element description used to obtain permission to interact with the element
119+
- `ref` (string): Exact target element reference from the page snapshot
120+
121+
- **browser_drag**
122+
- Description: Perform drag and drop between two elements
123+
- Parameters:
124+
- `startElement` (string): Human-readable source element description used to obtain permission to interact with the element
125+
- `startRef` (string): Exact source element reference from the page snapshot
126+
- `endElement` (string): Human-readable target element description used to obtain permission to interact with the element
127+
- `endRef` (string): Exact target element reference from the page snapshot
128+
129+
- **browser_type**
130+
- Description: Type text into editable element
131+
- Parameters:
132+
- `element` (string): Human-readable element description used to obtain permission to interact with the element
133+
- `ref` (string): Exact target element reference from the page snapshot
134+
- `text` (string): Text to type into the element
135+
- `submit` (boolean): Whether to submit entered text (press Enter after)
136+
137+
- **browser_press_key**
138+
- Description: Press a key on the keyboard
139+
- Parameters:
140+
- `key` (string): Name of the key to press or a character to generate, such as `ArrowLeft` or `a`
141+
142+
- **browser_snapshot**
143+
- Description: Capture accessibility snapshot of the current page (better than screenshot)
144+
- Parameters: None
145+
146+
- **browser_save_as_pdf**
147+
- Description: Save page as PDF
148+
- Parameters: None
149+
150+
- **browser_wait**
151+
- Description: Wait for a specified time in seconds
152+
- Parameters:
153+
- `time` (number): The time to wait in seconds (capped at 10 seconds)
154+
155+
- **browser_close**
156+
- Description: Close the page
157+
- Parameters: None
158+
159+
160+
### Vision Mode
161+
162+
Vision Mode provides tools for visual-based interactions using screenshots. Here are all available tools:
163+
164+
- **browser_navigate**
165+
- Description: Navigate to a URL
166+
- Parameters:
167+
- `url` (string): The URL to navigate to
168+
169+
- **browser_go_back**
170+
- Description: Go back to the previous page
171+
- Parameters: None
172+
173+
- **browser_go_forward**
174+
- Description: Go forward to the next page
175+
- Parameters: None
176+
177+
- **browser_screenshot**
178+
- Description: Capture screenshot of the current page
179+
- Parameters: None
180+
181+
- **browser_move_mouse**
182+
- Description: Move mouse to specified coordinates
183+
- Parameters:
184+
- `x` (number): X coordinate
185+
- `y` (number): Y coordinate
186+
187+
- **browser_click**
188+
- Description: Click at specified coordinates
189+
- Parameters:
190+
- `x` (number): X coordinate to click at
191+
- `y` (number): Y coordinate to click at
192+
193+
- **browser_drag**
194+
- Description: Perform drag and drop operation
195+
- Parameters:
196+
- `startX` (number): Start X coordinate
197+
- `startY` (number): Start Y coordinate
198+
- `endX` (number): End X coordinate
199+
- `endY` (number): End Y coordinate
200+
201+
- **browser_type**
202+
- Description: Type text at specified coordinates
203+
- Parameters:
204+
- `text` (string): Text to type
205+
- `submit` (boolean): Whether to submit entered text (press Enter after)
206+
207+
- **browser_press_key**
208+
- Description: Press a key on the keyboard
209+
- Parameters:
210+
- `key` (string): Name of the key to press or a character to generate, such as `ArrowLeft` or `a`
211+
212+
- **browser_save_as_pdf**
213+
- Description: Save page as PDF
214+
- Parameters: None
215+
216+
- **browser_wait**
217+
- Description: Wait for a specified time in seconds
218+
- Parameters:
219+
- `time` (number): The time to wait in seconds (capped at 10 seconds)
220+
221+
- **browser_close**
222+
- Description: Close the page
223+
- Parameters: None

‎cli.js

+18
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,18 @@
1+
#!/usr/bin/env node
2+
/**
3+
* Copyright (c) Microsoft Corporation.
4+
*
5+
* Licensed under the Apache License, Version 2.0 (the "License");
6+
* you may not use this file except in compliance with the License.
7+
* You may obtain a copy of the License at
8+
*
9+
* http://www.apache.org/licenses/LICENSE-2.0
10+
*
11+
* Unless required by applicable law or agreed to in writing, software
12+
* distributed under the License is distributed on an "AS IS" BASIS,
13+
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
14+
* See the License for the specific language governing permissions and
15+
* limitations under the License.
16+
*/
17+
18+
require('./lib/program');

0 commit comments

Comments
 (0)