Browser action engine. Provides up-to-date action manuals for the modern web — operate any website instantly, one tab or dozens, concurrently.
Detected risks:
Actionbook is a powerful browser action engine designed to automate and streamline interactions with modern websites. It solves the common problem of manually navigating, filling forms, clicking buttons, and extracting information from web pages. By providing up-to-date action manuals, Actionbook enables agents to operate websites with precision and speed, removing the need for repetitive trial-and-error navigation or element discovery. Its stateless architecture allows users to handle multiple sessions and tabs concurrently, making it ideal for complex workflows that require parallel operations across different sites or pages.
The main features of Actionbook include generating accurate action manuals that outline the page structure and element references, which can be used for filling forms, clicking buttons, navigating, and observing page changes. It offers a snapshot system that labels elements with stable references (e.g., @e3, @e7) to ensure commands remain consistent across interactions. Key capabilities include starting browser sessions, navigating to URLs, taking snapshots, performing actions based on element refs, waiting for page events, and running concurrent sessions and tabs. Commands are fully documented with integrated help, and the system supports automation for search, manual actions, session management, tab control, navigation, and observation tasks.
Actionbook is suitable for a wide range of users and use cases. Developers building browser-based AI agents, web scrapers, or end-to-end (E2E) tests will find it especially valuable for automating repetitive tasks like data entry, content posting, or monitoring page changes. Non-technical users can leverage Actionbook to perform actions across websites efficiently, such as sending messages, booking reservations, or interacting with dynamic content without manual effort. Its multi-platform support (macOS, Linux, Windows) and optional API key for unlimited requests make it accessible for both small-scale operations and large-scale web automation projects.
An API key is optional. Without it, the skill works but requests are subject to public rate limits. With a key, you can make unlimited requests. You can create one at https://actionbook.dev/dashboard.
Actionbook runs on macOS, Linux, and Windows, providing consistent functionality across these operating systems.
Yes. Actionbook's stateless architecture allows you to operate dozens of tabs and sessions concurrently by specifying session and tab IDs explicitly.
It uses snapshots to label elements with stable references. These refs remain consistent across snapshots as long as the element doesn't change, enabling reliable automation without re-snapshotting every step.
You can automate actions such as filling forms, clicking buttons, navigating, searching, filtering content, taking screenshots, monitoring changes, and more on virtually any website.
Quick Setup:
.claude/skills/Repository
actionbook/actionbook