Nova Act MCP Server

MCPOpen SourceMIT34.2

by madtank • Documentation & Docs

An MCP server that exposes Amazon Nova Act browser-automation tools to AI agents for browser control and inspection.

Example Use Cases

1
Start and manage real browser sessions to perform web automation and interaction via natural language instructions.
2
Capture page screenshots and page state (saved as files) for visual inspection or reporting.
3
Download files from web pages, collect HTML logs, and compress session artifacts for audit or analysis.

Description

This zero-install Model Context Protocol (MCP) server wraps Amazon Nova Act browser automation capabilities and exposes them as MCP tools (start_session, execute_instruction, inspect_browser, etc.). It enables agents to programmatically drive browsers, capture screenshots saved to files, fetch files from pages, and collect logs for offline analysis. The server is designed for easy integration with MCP-compatible clients and includes testing utilities and MCP Inspector support for development and validation.

Capabilities(9 total)

Tools (9)

start_session

Starts and initializes a new browser session controlled by Nova Act, navigating to the specified URL. Returns session details including a unique `session_id` required for other browser tools. This is a long-running operation (20-60s) and will send progress notifications.

url:string*headless:booleanidentity:stringsession_id:anyctx:any

execute_instruction

Executes a natural language instruction in an active browser session. Use `task` for the main instruction. Returns action outcome, URL, and agent thinking. For complex goals, call multiple times, verifying each step.

session_id:string*task:string*instructions:anytimeout:integerretry_attempts:integerquiet:booleanctx:any

inspect_browser

Retrieves current state (URL, title) of an active browser session. Screenshot is omitted by default. Set include_screenshot=True to get a base64 JPEG (can be large).

session_id:string*include_screenshot:boolean

end_session

Closes the specified browser session and cleans up resources. Call this when done with a session.

session_id:string*

fetch_file

Retrieves the contents of a file from the server, with optional base64 encoding. Primarily used to access log files, screenshots, and other assets generated during browser automation. Set encode_base64=True for binary files like images. Has a configurable size limit (default 10MB) to prevent excessive token usage with large files.

file_path:string*encode_base64:booleanmax_size:integer

Quick Actions

View on GitHub

Security

Scanned 3 month(s) ago

Risk Level

MINIMAL

Read-only data retrieval, no side effects

Trust Score

C52/100

5/17 checks passed

Scores are informational only and provided “as is” without warranty. AgentHotspot assumes no liability for actions taken based on these ratings.

Quick Stats

Service TypeMCP

Pricing ModelUnknown

Capabilities9 Tools / 0 Prompts / 0 Resources

Ownermadtank

CategoryDocumentation & Docs

DependenciesStandalone

Set Your Username