### Start MCP Browser Agent Server

Source: https://context7.com/imprvhub/mcp-browser-agent/llms.txt

Demonstrates how to start the MCP Browser Agent server using command-line arguments, a JSON configuration file, or environment variables. The server integrates with Claude Desktop via stdio transport.

```typescript
// src/index.ts — start the server manually
// node dist/index.js --browser chrome --viewport-width 1920 --viewport-height 1080 --device-scale-factor 1.0
```

```json
// ~/.mcp_browser_agent_config.json (auto-created/updated on start)
{
  "browserType": "chrome",
  "viewportWidth": 1920,
  "viewportHeight": 1080,
  "deviceScaleFactor": 1.0
}
```

```bash
// Environment variable alternative
// MCP_BROWSER_TYPE=firefox MCP_VIEWPORT_WIDTH=1280 node dist/index.js
```

```json
// Claude Desktop integration — claude_desktop_config.json
{
  "mcpServers": {
    "browserAgent": {
      "command": "node",
      "args": [
        "/absolute/path/to/mcp-browser-agent/dist/index.js",
        "--browser", "chrome",
        "--viewport-width", "1280",
        "--viewport-height", "800"
      ]
    }
  }
}
```

```text
// Expected: MCP server starts, connects via stdio, browser launches on first tool call
```

--------------------------------

### Clone and Install MCP Browser Agent

Source: https://github.com/imprvhub/mcp-browser-agent/blob/main/README.md

Steps to clone the MCP Browser Agent repository and install its dependencies using npm.

```bash
git clone https://github.com/imprvhub/mcp-browser-agent
cd mcp-browser-agent
npm install
```

--------------------------------

### Build MCP Browser Agent Project

Source: https://github.com/imprvhub/mcp-browser-agent/blob/main/README.md

Build the MCP Browser Agent project after installation.

```bash
npm run build
```

--------------------------------

### Configure Multiple MCP Servers in Claude Desktop

Source: https://github.com/imprvhub/mcp-browser-agent/blob/main/README.md

Example of configuring multiple MCP servers, including the browserAgent, within the Claude Desktop configuration file.

```json
{
  "mcpServers": {
    "otherMcp1": {
      "command": "...",
      "args": ["..."]
    },
    "otherMcp2": {
      "command": "...",
      "args": ["..."]
    },
    "browserAgent": {
      "command": "node",
      "args": [
        "ABSOLUTE_PATH_TO_DIRECTORY/mcp-browser-agent/dist/index.js",
      "--browser",
      "chrome"
    ]
    }
  }
}
```

--------------------------------

### Install Playwright Browsers

Source: https://github.com/imprvhub/mcp-browser-agent/blob/main/README.md

Manually install Playwright browser drivers for supported browsers. Playwright automatically installs drivers on first use, but these commands allow for manual installation.

```bash
npx playwright install chrome
npx playwright install firefox
npx playwright install webkit
npx playwright install msedge
```

--------------------------------

### Run MCP Server with Firefox Argument

Source: https://github.com/imprvhub/mcp-browser-agent/blob/main/README.md

Start the MCP server manually and specify Firefox as the browser type using a command-line argument.

```bash
node dist/index.js --browser firefox
```

--------------------------------

### MCP Tool Call: browser_navigate

Source: https://context7.com/imprvhub/mcp-browser-agent/llms.txt

Illustrates an MCP tool call to navigate the browser to a specified URL. Includes optional parameters for timeout and `waitUntil` strategies, and shows example success and error responses.

```json
// MCP tool call (as Claude would issue it)
{
  "name": "browser_navigate",
  "arguments": {
    "url": "https://example.com",
    "timeout": 15000,
    "waitUntil": "networkidle"
  }
}
```

```json
// Success response:
// { "content": [{ "type": "text", "text": "Navigated to https://example.com" }], "isError": false }
```

```json
// Error response (unreachable host):
// { "content": [{ "type": "text", "text": "Navigation failed: net::ERR_NAME_NOT_RESOLVED" }], "isError": true }
```

--------------------------------

### Configure Claude Desktop for Browser Agent

Source: https://github.com/imprvhub/mcp-browser-agent/blob/main/README.md

Add the Browser Agent MCP configuration to the Claude Desktop configuration file to enable auto-starting. Replace ABSOLUTE_PATH_TO_DIRECTORY with the actual path to your mcp-browser-agent installation.

```json
{
  "mcpServers": {
    "browserAgent": {
      "command": "node",
      "args": ["ABSOLUTE_PATH_TO_DIRECTORY/mcp-browser-agent/dist/index.js",
      "--browser",
      "chrome"
    ]
    }
  }
}
```

--------------------------------

### Perform GET Request with API Tool

Source: https://github.com/imprvhub/mcp-browser-agent/blob/main/README.md

Use the `api_get` tool to make an HTTP GET request to a specified URL. Optional `headers` can be provided.

```text
Perform a GET request to https://jsonplaceholder.typicode.com/todos/1
```

--------------------------------

### Run MCP Server with Firefox Environment Variable

Source: https://github.com/imprvhub/mcp-browser-agent/blob/main/README.md

Start the MCP server manually and specify Firefox as the browser type using the MCP_BROWSER_TYPE environment variable.

```bash
MCP_BROWSER_TYPE=firefox node dist/index.js
```

--------------------------------

### Perform HTTP GET Request with api_get

Source: https://context7.com/imprvhub/mcp-browser-agent/llms.txt

Performs an HTTP GET request and returns the status code and formatted response body. JSON responses are pretty-printed; other content types are returned as raw text. Use for fetching data from APIs.

```json
{
  "name": "api_get",
  "arguments": {
    "url": "https://jsonplaceholder.typicode.com/todos/1",
    "headers": { "Accept": "application/json" }
  }
}
```

--------------------------------

### api_get

Source: https://context7.com/imprvhub/mcp-browser-agent/llms.txt

Performs an HTTP GET request and returns the status code and formatted response body. JSON responses are pretty-printed; other content types are returned as raw text.

```APIDOC
## api_get — HTTP GET Request

Performs an HTTP GET request and returns the status code and formatted response body. JSON responses are pretty-printed; other content types are returned as raw text.

### Arguments
* `url` (string): The URL to send the GET request to.
* `headers` (object, optional): An object containing request headers.

### Example
```json
{
  "name": "api_get",
  "arguments": {
    "url": "https://jsonplaceholder.typicode.com/todos/1",
    "headers": { "Accept": "application/json" }
  }
}
```

### Response
```json
{
  "content": [
    { "type": "text", "text": "GET https://jsonplaceholder.typicode.com/todos/1 - Status: 200" },
    { "type": "text", "text": "Response body:\n{\n  \"userId\": 1,\n  \"id\": 1,\n  \"title\": \"delectus aut autem\",\n  \"completed\": false\n}" }
  ], "isError": false
}
```
```

--------------------------------

### Initialize Playwright Project

Source: https://github.com/imprvhub/mcp-browser-agent/blob/main/README.md

Use this command to initialize a new Playwright project, which includes Playwright and necessary dependencies for browser automation.

```bash
npm init playwright@latest
```

--------------------------------

### Login Form Interaction with Browser Tools

Source: https://github.com/imprvhub/mcp-browser-agent/blob/main/README.md

Demonstrates filling form fields for username and password, followed by clicking the login button. Requires selectors for the input fields and the button.

```text
Navigate to https://the-internet.herokuapp.com/login and fill in the username field with "tomsmith" and the password field with "SuperSecretPassword!"
```

```text
Go to https://the-internet.herokuapp.com/login, fill in the username and password fields, then click the login button
```

--------------------------------

### Navigate and Search with Browser Tool

Source: https://github.com/imprvhub/mcp-browser-agent/blob/main/README.md

Combines navigation and interaction by first navigating to a URL and then performing a search action. Assumes the search input is identifiable by a selector.

```text
Navigate to https://www.wikipedia.org and search for "Model Context Protocol"
```

--------------------------------

### Run Tests

Source: https://github.com/imprvhub/mcp-browser-agent/blob/main/README.md

Execute the project's test suite. Use watch mode for continuous testing or coverage for a detailed report.

```bash
npm test               # Run tests
npm run test:watch     # Watch mode
npm run test:coverage  # Coverage report
```

--------------------------------

### Navigate to URL with Browser Tool

Source: https://github.com/imprvhub/mcp-browser-agent/blob/main/README.md

Use the `browser_navigate` tool to direct the browser to a specified URL. Supports optional `timeout` and `waitUntil` parameters.

```text
Navigate to the Google homepage at https://www.google.com
```

--------------------------------

### Register MCP Browser and API Tools

Source: https://context7.com/imprvhub/mcp-browser-agent/llms.txt

Shows how to register all available browser and API tools using `registerTools()` from the MCP SDK. The registered tools are exposed to Claude via the `tools/list` protocol method.

```typescript
import { registerTools, BROWSER_TOOLS, API_TOOLS } from "./tools.js";

const tools = registerTools();
// tools is Tool[] with 12 entries:
// BROWSER_TOOLS: ["browser_set_viewport","browser_navigate","browser_screenshot",
//                 "browser_click","browser_fill","browser_select","browser_hover","browser_evaluate"]
// API_TOOLS:     ["api_get","api_post","api_put","api_patch","api_delete"]

console.log(tools.map(t => t.name));
```

```text
// Output:
// ["browser_set_viewport","browser_navigate","browser_screenshot","browser_click",
//  "browser_fill","browser_select","browser_hover","browser_evaluate",
//  "api_get","api_post","api_put","api_patch","api_delete"]
```

--------------------------------

### Watch for Changes

Source: https://github.com/imprvhub/mcp-browser-agent/blob/main/README.md

Use this command to watch for file changes during development.

```bash
npm run watch
```

--------------------------------

### Configure Browser Type via .mcp_browser_agent_config.json

Source: https://github.com/imprvhub/mcp-browser-agent/blob/main/README.md

Specify the browser type for the MCP Browser Agent by creating or editing the .mcp_browser_agent_config.json file in your home directory.

```json
{
  "browserType": "chrome"
}
```

--------------------------------

### Select a Dropdown Option with browser_select

Source: https://context7.com/imprvhub/mcp-browser-agent/llms.txt

Waits for a `<select>` element and selects the matching option by value or label text. Ensure the selector targets a valid dropdown element.

```json
{
  "name": "browser_select",
  "arguments": {
    "selector": "#dropdown",
    "value": "Option 2"
  }
}
```

--------------------------------

### Access MCP Resources: Logs and Screenshots

Source: https://context7.com/imprvhub/mcp-browser-agent/llms.txt

Access browser console logs via `browser://logs` and previously captured screenshots via `screenshot://<name>`. Requires the MCP SDK client.

```typescript
import { Client } from "@modelcontextprotocol/sdk/client/index.js";

// List all available resources
const { resources } = await client.listResources();
// [ 
//   { uri: "browser://logs",       mimeType: "text/plain", name: "Browser console logs" },
//   { uri: "screenshot://homepage",mimeType: "image/png",  name: "Screenshot: homepage" } 
// ]

// Read console logs
const logsResource = await client.readResource({ uri: "browser://logs" });
console.log(logsResource.contents[0].text);
// "[log] title: Example Domain\n[warn] Deprecated API called"

// Read a screenshot by name
const screenshotResource = await client.readResource({ uri: "screenshot://homepage" });
const base64Data = screenshotResource.contents[0].blob;
// base64-encoded PNG of the "homepage" screenshot
```

--------------------------------

### Configure Browser Agent in Claude Desktop Config

Source: https://github.com/imprvhub/mcp-browser-agent/blob/main/README.md

Specify the browser type for the Browser Agent within the Claude Desktop configuration file. Replace ABSOLUTE_PATH_TO_DIRECTORY with the actual path.

```json
{
  "mcpServers": {
    "browserAgent": {
      "command": "node",
      "args": [
        "ABSOLUTE_PATH_TO_DIRECTORY/mcp-browser-agent/dist/index.js",
        "--browser",
        "chrome"
      ]
    }
  }
}
```

--------------------------------

### MCP Resources

Source: https://context7.com/imprvhub/mcp-browser-agent/llms.txt

Access MCP resources like browser logs and screenshots.

```APIDOC
## MCP Resources

### Description
The server exposes two MCP resource types: `browser://logs` for console messages and `screenshot://<name>` for captured screenshots.

### Accessing Resources (SDK Example)
```typescript
import { Client } from "@modelcontextprotocol/sdk/client/index.js";

const client = new Client();

// List all available resources
const { resources } = await client.listResources();
console.log(resources);
// [ 
//   { uri: "browser://logs", mimeType: "text/plain", name: "Browser console logs" },
//   { uri: "screenshot://homepage", mimeType: "image/png", name: "Screenshot: homepage" } 
// ]

// Read console logs
const logsResource = await client.readResource({ uri: "browser://logs" });
console.log(logsResource.contents[0].text);
// "[log] title: Example Domain\n[warn] Deprecated API called"

// Read a screenshot by name
const screenshotResource = await client.readResource({ uri: "screenshot://homepage" });
const base64Data = screenshotResource.contents[0].blob;
// base64-encoded PNG of the "homepage" screenshot
```
```

--------------------------------

### Select Dropdown Option with Browser Tool

Source: https://github.com/imprvhub/mcp-browser-agent/blob/main/README.md

The `browser_select` tool allows for selecting a specific option from a dropdown element. Requires `selector` and `value` parameters.

```text
Go to https://the-internet.herokuapp.com/dropdown and select the option "Option 1" from the dropdown
```

--------------------------------

### browser_select

Source: https://context7.com/imprvhub/mcp-browser-agent/llms.txt

Waits for a `<select>` element and selects the matching option by value or label text.

```APIDOC
## browser_select — Select a Dropdown Option

Waits for a `<select>` element and selects the matching option by value or label text.

### Arguments
* `selector` (string): The CSS selector for the select element.
* `value` (string): The value or label text of the option to select.

### Example
```json
{
  "name": "browser_select",
  "arguments": {
    "selector": "#dropdown",
    "value": "Option 2"
  }
}
```

### Response
* Success: `{"content": [{"type": "text", "text": "Selected option Option 2 in #dropdown"}], "isError": false}`
* Error: `{"content": [{"type": "text", "text": "Selection failed on #dropdown: ..."}], "isError": true}`
```

--------------------------------

### MCP Tool Call: browser_screenshot

Source: https://context7.com/imprvhub/mcp-browser-agent/llms.txt

Demonstrates capturing screenshots using the `browser_screenshot` tool. Supports full-page or element-specific screenshots, with options for naming, saving to a path, and masking specific elements.

```json
// Full-page screenshot
{
  "name": "browser_screenshot",
  "arguments": {
    "name": "homepage",
    "fullPage": true,
    "savePath": "/tmp/screenshots"
  }
}
```

```json
// Response:
// { "content": [
//     { "type": "text",  "text": "Screenshot saved to: /tmp/screenshots/homepage-2024-05-01T12-00-00-000Z.png" },
//     { "type": "image", "data": "<base64>", "mimeType": "image/png" }
//   ], "isError": false }
```

```json
// Element-specific screenshot with masking
{
  "name": "browser_screenshot",
  "arguments": {
    "name": "header-only",
    "selector": "header.site-header",
    "mask": [".cookie-banner", "#ads"]
  }
}
```

```text
// Screenshot of the matched element with masked regions blurred out
```

--------------------------------

### Run MCP Server Manually

Source: https://github.com/imprvhub/mcp-browser-agent/blob/main/README.md

Execute the MCP server directly using Node.js. Keep this terminal open while using Claude Desktop.

```bash
node dist/index.js
```

--------------------------------

### browser_screenshot

Source: https://context7.com/imprvhub/mcp-browser-agent/llms.txt

Takes a PNG screenshot of the full page or a specific CSS-selected element. The image is returned as a base64-encoded MCP image content block.

```APIDOC
## browser_screenshot

### Description
Takes a PNG screenshot of the full page or a specific CSS-selected element. The image is saved to a specified path and returned as a base64-encoded MCP image content block.

### Method
POST (MCP Tool Call)

### Endpoint
N/A (MCP Tool Call)

### Parameters
#### Arguments
- **name** (string) - Required - A name for the screenshot.
- **fullPage** (boolean) - Optional - If true, captures the full page. Defaults to false.
- **savePath** (string) - Optional - The path to save the screenshot file. If not provided, it defaults to the user's Downloads folder.
- **selector** (string) - Optional - A CSS selector to capture a specific element. If provided, `fullPage` is ignored.
- **mask** (array of strings) - Optional - An array of CSS selectors for regions to mask (blur) in the screenshot.

### Request Example
```json
// Full-page screenshot
{
  "name": "homepage",
  "arguments": {
    "fullPage": true,
    "savePath": "/tmp/screenshots"
  }
}

// Element-specific screenshot with masking
{
  "name": "header-only",
  "arguments": {
    "selector": "header.site-header",
    "mask": [".cookie-banner", "#ads"]
  }
}
```

### Response
#### Success Response (200)
- **content** (array) - An array containing text about the save location and an image content block.
- **isError** (boolean) - False if the operation was successful.

#### Response Example
```json
{
  "content": [
    { "type": "text",  "text": "Screenshot saved to: /tmp/screenshots/homepage-2024-05-01T12-00-00-000Z.png" },
    { "type": "image", "data": "<base64>", "mimeType": "image/png" }
  ],
  "isError": false
}
```
```

--------------------------------

### Fill Form Input with Browser Tool

Source: https://github.com/imprvhub/mcp-browser-agent/blob/main/README.md

Use the `browser_fill` tool to input text into a form field identified by a selector. Both `selector` and `value` are required.

```text
Type "weather forecast" in the search box
```

--------------------------------

### browser_navigate

Source: https://context7.com/imprvhub/mcp-browser-agent/llms.txt

Navigates the persistent browser page to the specified URL. Supports configurable timeout and Playwright `waitUntil` strategies.

```APIDOC
## browser_navigate

### Description
Navigates the persistent browser page to the specified URL. Supports configurable timeout and Playwright `waitUntil` strategies (`load`, `domcontentloaded`, `networkidle`, `commit`).

### Method
POST (MCP Tool Call)

### Endpoint
N/A (MCP Tool Call)

### Parameters
#### Arguments
- **url** (string) - Required - The URL to navigate to.
- **timeout** (number) - Optional - The navigation timeout in milliseconds.
- **waitUntil** (string) - Optional - The Playwright `waitUntil` strategy. Possible values: `load`, `domcontentloaded`, `networkidle`, `commit`.

### Request Example
```json
{
  "name": "browser_navigate",
  "arguments": {
    "url": "https://example.com",
    "timeout": 15000,
    "waitUntil": "networkidle"
  }
}
```

### Response
#### Success Response (200)
- **content** (array) - An array of content blocks, typically text indicating success.
- **isError** (boolean) - False if the operation was successful.

#### Response Example
```json
{
  "content": [{ "type": "text", "text": "Navigated to https://example.com" }],
  "isError": false
}
```

#### Error Response Example
```json
{
  "content": [{ "type": "text", "text": "Navigation failed: net::ERR_NAME_NOT_RESOLVED" }],
  "isError": true
}
```
```

--------------------------------

### Change Browser Viewport with browser_set_viewport

Source: https://context7.com/imprvhub/mcp-browser-agent/llms.txt

Dynamically resizes the browser viewport and optionally adjusts the device scale factor. The new dimensions are persisted for future sessions. Useful for testing responsive designs.

```json
{
  "name": "browser_set_viewport",
  "arguments": {
    "width": 375,
    "height": 812,
    "deviceScaleFactor": 3.0
  }
}
```

--------------------------------

### Perform POST Request with API Tool

Source: https://github.com/imprvhub/mcp-browser-agent/blob/main/README.md

The `api_post` tool sends an HTTP POST request to a URL with required `data` and optional `headers`.

```text
Make a POST request to https://jsonplaceholder.typicode.com/posts with appropriate JSON data
```

--------------------------------

### Execute JavaScript with Browser Tool

Source: https://github.com/imprvhub/mcp-browser-agent/blob/main/README.md

The `browser_evaluate` tool executes arbitrary JavaScript code within the browser context. The script should return the desired value.

```text
Go to https://example.com and execute a JavaScript script to return the page title
```

```text
Navigate to https://www.google.com and execute a JavaScript script to count the number of links on the page
```

--------------------------------

### browser_set_viewport

Source: https://context7.com/imprvhub/mcp-browser-agent/llms.txt

Dynamically resizes the browser viewport and optionally adjusts the device scale factor. The new dimensions are persisted to `~/.mcp_browser_agent_config.json` for future sessions.

```APIDOC
## browser_set_viewport — Change Browser Viewport

Dynamically resizes the browser viewport and optionally adjusts the device scale factor. The new dimensions are persisted to `~/.mcp_browser_agent_config.json` for future sessions.

### Arguments
* `width` (integer): The desired width of the viewport.
* `height` (integer): The desired height of the viewport.
* `deviceScaleFactor` (float, optional): The device scale factor.

### Example
```json
{
  "name": "browser_set_viewport",
  "arguments": {
    "width": 375,
    "height": 812,
    "deviceScaleFactor": 3.0
  }
}
```

### Use Case
Simulate iPhone 12 Pro screen for mobile layout testing.

### Response
* Success: `{"content": [{"type": "text", "text": "Set viewport to width: 375, height: 812, scale factor: 3"}], "isError": false}`
```

--------------------------------

### executeToolCall

Source: https://context7.com/imprvhub/mcp-browser-agent/llms.txt

The central dispatch function for executing tools, lazily initializing clients and routing to handlers.

```APIDOC
## executeToolCall()

### Description
`executeToolCall(toolName, args, server)` is the central dispatch function. It lazily initializes the browser or API client as needed, routes to the appropriate handler, and returns a `{ toolResult: CallToolResult }` object. All errors are caught and returned as non-throwing `isError: true` results.

### Method Signature
`executeToolCall(toolName: string, args: object, server: object): Promise<{ toolResult: CallToolResult }> `

### Parameters
- **toolName** (string) - The name of the tool to execute.
- **args** (object) - The arguments to pass to the tool.
- **server** (object) - The server instance.

### Usage Example
```typescript
import { executeToolCall } from "./executor.js";

// Direct programmatic invocation (e.g., in tests or custom integrations)
const result = await executeToolCall(
  "browser_navigate",
  { url: "https://example.com", waitUntil: "load" },
  server
);

if (result.toolResult.isError) {
  console.error("Tool failed:", result.toolResult.content[0].text);
} else {
  console.log("Success:", result.toolResult.content[0].text);
  // "Navigated to https://example.com"
}

// Unknown tool gracefully returns an error
const unknown = await executeToolCall("nonexistent_tool", {}, server);
console.log(unknown.toolResult.isError);   // true
console.log(unknown.toolResult.content[0].text); // "Unknown tool: nonexistent_tool"
```

### Response
Returns an object containing `toolResult`. `toolResult` has an `isError` boolean and a `content` array. If `isError` is true, `content` contains error details. Otherwise, it contains the tool's successful output.
```

--------------------------------

### Capture Screenshot with Browser Tool

Source: https://github.com/imprvhub/mcp-browser-agent/blob/main/README.md

Utilize the `browser_screenshot` tool to capture a screenshot of the current page or a specific element. Requires a `name` and supports `selector`, `fullPage`, `mask`, and `savePath` options.

```text
Take a screenshot of the current page and name it "google-homepage"
```

--------------------------------

### Dispatch Tool Calls with executeToolCall

Source: https://context7.com/imprvhub/mcp-browser-agent/llms.txt

Use `executeToolCall` to invoke browser or API clients. It handles initialization, routing, and error catching, returning results as `{ toolResult: CallToolResult }`.

```typescript
import { executeToolCall } from "./executor.js";

// Direct programmatic invocation (e.g., in tests or custom integrations)
const result = await executeToolCall(
  "browser_navigate",
  { url: "https://example.com", waitUntil: "load" },
  server
);

if (result.toolResult.isError) {
  console.error("Tool failed:", result.toolResult.content[0].text);
} else {
  console.log("Success:", result.toolResult.content[0].text);
  // "Navigated to https://example.com"
}

// Unknown tool gracefully returns an error
const unknown = await executeToolCall("nonexistent_tool", {}, server);
console.log(unknown.toolResult.isError);   // true
console.log(unknown.toolResult.content[0].text); // "Unknown tool: nonexistent_tool"
```

--------------------------------

### Fill a Form Input with browser_fill

Source: https://context7.com/imprvhub/mcp-browser-agent/llms.txt

Waits for the selector to appear, then fills it with the provided text value. Suitable for text inputs, textareas, and contenteditable elements. Can be chained with other commands.

```json
{
  "name": "browser_fill",
  "arguments": {
    "selector": "input#username",
    "value": "tomsmith"
  }
}
```

--------------------------------

### api_post

Source: https://context7.com/imprvhub/mcp-browser-agent/llms.txt

Sends an HTTP POST request with a JSON body. Defaults `Content-Type` to `application/json` unless overridden via headers.

```APIDOC
## api_post — HTTP POST Request

Sends an HTTP POST request with a JSON body. Defaults `Content-Type` to `application/json` unless overridden via headers.

### Arguments
* `url` (string): The URL to send the POST request to.
* `data` (string): The JSON string to send as the request body.
* `headers` (object, optional): An object containing request headers. Defaults to `{"Content-Type": "application/json"}` if not provided.

### Example
```json
{
  "name": "api_post",
  "arguments": {
    "url": "https://jsonplaceholder.typicode.com/posts",
    "data": "{\"title\": \"foo\", \"body\": \"bar\", \"userId\": 1}",
    "headers": { "Content-Type": "application/json" }
  }
}
```

### Response
```json
{
  "content": [
    { "type": "text", "text": "POST https://jsonplaceholder.typicode.com/posts - Status: 201" },
    { "type": "text", "text": "Response body:\n{\n  \"title\": \"foo\",\n  \"body\": \"bar\",\n  \"userId\": 1,\n  \"id\": 101\n}" }
  ], "isError": false
}
```
```

--------------------------------

### browser_click

Source: https://context7.com/imprvhub/mcp-browser-agent/llms.txt

Clicks the element matched by the given CSS selector on the current page.

```APIDOC
## browser_click — Click a DOM Element

Clicks the element matched by the given CSS selector on the current page.

### Arguments
* `selector` (string): The CSS selector for the element to click.

### Example
```json
{
  "name": "browser_click",
  "arguments": {
    "selector": "button[type='submit']"
  }
}
```

### Response
* Success: `{"content": [{"type": "text", "text": "Clicked element: button[type='submit']"}], "isError": false}`
* Error: `{"content": [{"type": "text", "text": "Click failed on button[type='submit']: Element not found"}], "isError": true}`
```

--------------------------------

### browser_fill

Source: https://context7.com/imprvhub/mcp-browser-agent/llms.txt

Waits for the selector to appear, then fills it with the provided text value. Suitable for text inputs, textareas, and contenteditable elements.

```APIDOC
## browser_fill — Fill a Form Input

Waits for the selector to appear, then fills it with the provided text value. Suitable for text inputs, textareas, and contenteditable elements.

### Arguments
* `selector` (string): The CSS selector for the input element.
* `value` (string): The text value to fill into the input.

### Example
```json
{
  "name": "browser_fill",
  "arguments": {
    "selector": "input#username",
    "value": "tomsmith"
  }
}
```

### Chaining Example
```
// Chain with password fill and submit:
browser_fill { selector: "input#password", value: "SuperSecretPassword!" }
browser_click { selector: "button[type='submit']" }
```
```

--------------------------------

### Execute JavaScript in the Browser with browser_evaluate

Source: https://context7.com/imprvhub/mcp-browser-agent/llms.txt

Evaluates arbitrary JavaScript in the page context using `page.evaluate()`. Captures `console.log/info/warn/error` output alongside the return value. Useful for extracting data or triggering client-side logic.

```json
{
  "name": "browser_evaluate",
  "arguments": {
    "script": "document.querySelectorAll('a').length"
  }
}
```

```json
{
  "name": "browser_evaluate",
  "arguments": {
    "script": "console.log('title:', document.title); document.title"
  }
}
```

--------------------------------

### api_put

Source: https://context7.com/imprvhub/mcp-browser-agent/llms.txt

Replaces an existing resource with the supplied JSON body using an HTTP PUT request.

```APIDOC
## PUT /posts/{id}

### Description
Replaces an existing resource with the supplied JSON body.

### Method
PUT

### Endpoint
https://jsonplaceholder.typicode.com/posts/1

### Request Body
- **id** (integer) - Required - The ID of the post to update.
- **title** (string) - Required - The new title for the post.
- **body** (string) - Required - The new content for the post.
- **userId** (integer) - Required - The ID of the user who owns the post.

### Request Example
```json
{
  "id": 1,
  "title": "updated title",
  "body": "updated body",
  "userId": 1
}
```

### Response
#### Success Response (200)
- **id** (integer) - The ID of the updated post.
- **title** (string) - The updated title of the post.
- **body** (string) - The updated content of the post.
- **userId** (integer) - The ID of the user who owns the post.

#### Response Example
```json
{
  "id": 1,
  "title": "updated title",
  "body": "updated body",
  "userId": 1
}
```
```

--------------------------------

### Click a DOM Element with browser_click

Source: https://context7.com/imprvhub/mcp-browser-agent/llms.txt

Clicks the element matched by the given CSS selector on the current page. Use this for simulating user clicks on buttons, links, or other interactive elements.

```json
{
  "name": "browser_click",
  "arguments": {
    "selector": "button[type='submit']"
  }
}
```

--------------------------------

### browser_evaluate

Source: https://context7.com/imprvhub/mcp-browser-agent/llms.txt

Evaluates arbitrary JavaScript in the page context using `page.evaluate()`. Captures `console.log/info/warn/error` output alongside the return value.

```APIDOC
## browser_evaluate — Execute JavaScript in the Browser

Evaluates arbitrary JavaScript in the page context using `page.evaluate()`. Captures `console.log/info/warn/error` output alongside the return value.

### Arguments
* `script` (string): The JavaScript code to execute.

### Example 1: Get element count
```json
{
  "name": "browser_evaluate",
  "arguments": {
    "script": "document.querySelectorAll('a').length"
  }
}
```

### Response 1
```json
{
  "content": [
    { "type": "text", "text": "Script result: 42" },
    { "type": "text", "text": "Console output:\n" }
  ], "isError": false
}
```

### Example 2: Script with console output
```json
{
  "name": "browser_evaluate",
  "arguments": {
    "script": "console.log('title:', document.title); document.title"
  }
}
```

### Response 2
```json
{
  "content": [
    { "type": "text", "text": "Script result: \"Example Domain\"" },
    { "type": "text", "text": "Console output:\n[log] title: Example Domain" }
  ], "isError": false
}
```
```

--------------------------------

### Perform HTTP PUT Request with api_put

Source: https://context7.com/imprvhub/mcp-browser-agent/llms.txt

Use `api_put` to replace an existing resource with a JSON body. Ensure the URL and data are correctly formatted.

```json
{
  "name": "api_put",
  "arguments": {
    "url": "https://jsonplaceholder.typicode.com/posts/1",
    "data": "{\"id\": 1, \"title\": \"updated title\", \"body\": \"updated body\", \"userId\": 1}"
  }
}
```

--------------------------------

### Perform HTTP POST Request with api_post

Source: https://context7.com/imprvhub/mcp-browser-agent/llms.txt

Sends an HTTP POST request with a JSON body. Defaults `Content-Type` to `application/json` unless overridden via headers. Use for submitting data to APIs.

```json
{
  "name": "api_post",
  "arguments": {
    "url": "https://jsonplaceholder.typicode.com/posts",
    "data": "{\"title\": \"foo\", \"body\": \"bar\", \"userId\": 1}",
    "headers": { "Content-Type": "application/json" }
  }
}
```

--------------------------------

### browser_hover

Source: https://context7.com/imprvhub/mcp-browser-agent/llms.txt

Moves the mouse pointer over the element matched by the CSS selector, triggering CSS `:hover` states and JavaScript `mouseover` / `mouseenter` events.

```APIDOC
## browser_hover — Hover Over an Element

Moves the mouse pointer over the element matched by the CSS selector, triggering CSS `:hover` states and JavaScript `mouseover` / `mouseenter` events.

### Arguments
* `selector` (string): The CSS selector for the element to hover over.

### Example
```json
{
  "name": "browser_hover",
  "arguments": {
    "selector": "nav .dropdown-trigger"
  }
}
```

### Response
* Success: `{"content": [{"type": "text", "text": "Hovered over element: nav .dropdown-trigger"}], "isError": false}`
```

--------------------------------

### api_delete

Source: https://context7.com/imprvhub/mcp-browser-agent/llms.txt

Sends an HTTP DELETE request and returns the response status. No response body is included for delete operations.

```APIDOC
## DELETE /posts/{id}

### Description
Sends an HTTP DELETE request and returns the response status. No response body is included for delete operations.

### Method
DELETE

### Endpoint
https://jsonplaceholder.typicode.com/posts/1

### Headers
- **Authorization** (string) - Required - Bearer token for authentication.

### Response
#### Success Response (200)
Indicates successful deletion. No response body is returned.
```

--------------------------------

### Perform HTTP DELETE Request with api_delete

Source: https://context7.com/imprvhub/mcp-browser-agent/llms.txt

Use `api_delete` to send an HTTP DELETE request. This function returns the response status and does not include a response body.

```json
{
  "name": "api_delete",
  "arguments": {
    "url": "https://jsonplaceholder.typicode.com/posts/1",
    "headers": { "Authorization": "Bearer my-token" }
  }
}
```

--------------------------------

### Perform HTTP PATCH Request with api_patch

Source: https://context7.com/imprvhub/mcp-browser-agent/llms.txt

Use `api_patch` for partial updates to a resource. Only fields provided in the `data` argument will be modified.

```json
{
  "name": "api_patch",
  "arguments": {
    "url": "https://jsonplaceholder.typicode.com/posts/1",
    "data": "{\"title\": \"patched title\"}"
  }
}
```

--------------------------------

### Hover Over an Element with browser_hover

Source: https://context7.com/imprvhub/mcp-browser-agent/llms.txt

Moves the mouse pointer over the element matched by the CSS selector. This triggers CSS `:hover` states and JavaScript `mouseover` / `mouseenter` events.

```json
{
  "name": "browser_hover",
  "arguments": {
    "selector": "nav .dropdown-trigger"
  }
}
```

--------------------------------

### api_patch

Source: https://context7.com/imprvhub/mcp-browser-agent/llms.txt

Partially updates a resource using an HTTP PATCH request. Only the fields supplied in the data are modified.

```APIDOC
## PATCH /posts/{id}

### Description
Partially updates a resource. Only the fields supplied in `data` are modified.

### Method
PATCH

### Endpoint
https://jsonplaceholder.typicode.com/posts/1

### Request Body
- **title** (string) - Optional - The new title for the post.
- **body** (string) - Optional - The new content for the post.

### Request Example
```json
{
  "title": "patched title"
}
```

### Response
#### Success Response (200)
- **id** (integer) - The ID of the updated post.
- **title** (string) - The updated title of the post.
- **body** (string) - The updated content of the post.
- **userId** (integer) - The ID of the user who owns the post.

#### Response Example
```json
{
  "id": 1,
  "title": "patched title",
  "body": "updated body",
  "userId": 1
}
```
```

=== COMPLETE CONTENT === This response contains all available snippets from this library. No additional content exists. Do not make further requests.