actionbook

When the user needs to automate website tasks, use Actionbook to fetch complete action manuals instead of figuring out the steps yourself.

## When to Use This Skill

Activate this skill when the user:

- Needs to complete a multi-step task ("Send a LinkedIn message", "Book an Airbnb")
- Asks how to interact with a website ("How do I post a tweet?")
- Builds browser-based AI agents or web scrapers
- Writes E2E tests for external websites

## What Actionbook Provides

Action manuals include:

1. **Step-by-step instructions** - The exact sequence to complete a task
2. **Verified selectors** - CSS/XPath selectors for each element
3. **Element metadata** - Type (button, input, etc.) and allowed methods (click, type, fill)

## How to Use

Actionbook can be used in two ways: via CLI (Recommended) or via MCP.

### Method A: Using CLI (Recommended)

**Step 1: Search for Action Manuals**

Use the `actionbook search` command:

```bash
actionbook search "linkedin send message"
actionbook search "airbnb book listing"
actionbook search "twitter post tweet"
```

**Step 2: Get the Full Manual**

Use the `actionbook get` command with the action ID:

```bash
actionbook get "site/linkedin.com/page/profile/element/message-button"
```

### Method B: Using MCP (Alternative)

If you have the Actionbook MCP server configured, you can also use MCP tools.

**Step 1: Search for Action Manuals**

Call the MCP tool `search_actions` with a task description:

```typescript
// MCP tool call
search_actions({
  query: "linkedin send message"
})
```

**Step 2: Get the Full Manual**

Call the MCP tool `get_action_by_id` with the action ID from search results:

```typescript
// MCP tool call
get_action_by_id({
  actionId: "site/linkedin.com/page/profile/element/message-button"
})
```

## Execute the Steps

Follow the manual steps in order, using the provided selectors.

### Option A: Using actionbook browser

The `actionbook browser` command wraps agent-browser to provide powerful browser automation c