Zapier Plasmate Integration

Connect Plasmate to 7,000+ apps with Zapier. Plasmate is the browser engine for AI agents - it converts HTML to structured JSON (SOM) with 10-100x token compression.

Features

• Fetch Page - Convert any webpage to structured JSON, text, or markdown
• Extract Text - Get clean, readable text from any URL
• Extract Links - Pull all links from a page with filtering options
• Batch Fetch - Process up to 20 URLs in a single action
• Page Changed Trigger - Monitor webpages for content changes

Installation

Prerequisites

• Node.js 18+
• Zapier CLI
• A Zapier developer account

Setup

# Clone the repository
git clone https://github.com/nicholasrubright/zapier-plasmate.git
cd zapier-plasmate

# Install dependencies
npm install

# Login to Zapier CLI
zapier login

# Validate the integration
zapier validate

# Run tests
npm test

Development

Local Testing

# Run all tests
npm test

# Validate Zapier integration
zapier validate

# Test a specific action
zapier invoke create fetchPage --inputData '{"url": "https://example.com"}'

Environment Variables

For testing with Plasmate Cloud:

export PLASMATE_API_KEY="your-api-key"
export PLASMATE_BASE_URL="https://api.plasmate.app"

For local Plasmate:

export PLASMATE_BASE_URL="http://localhost:3000"

Publishing

Initial Publish

# Register the app with Zapier
zapier register "Plasmate"

# Push to Zapier
zapier push

# Promote to public (after Zapier review)
zapier promote 1.0.0

Updates

# Bump version in package.json, then:
zapier push

# Promote the new version
zapier promote 1.0.1

Actions

Fetch Page

Fetches a webpage and converts it to structured JSON (SOM) with massive token compression.

Inputs:

Field	Type	Required	Description
URL	string	Yes	The webpage URL to fetch
Format	choice	No	som (default), text, or markdown
CSS Selector	string	No	Extract specific content (e.g., article, .main)
Timeout	integer	No	Max wait time in ms (default: 30000)
Include Metadata	boolean	No	Include title, description, etc.
Include Links	boolean	No	Include all page links

Outputs:

• content - The processed content
• tokenSavings - Compression percentage (e.g., "85%")
• title, description - Page metadata
• links, linkCount - Extracted links (if enabled)

Extract Text

Extracts clean, readable text from a webpage.

Inputs:

Field	Type	Required	Description
URL	string	Yes	The webpage URL
CSS Selector	string	No	Extract from specific element
Max Length	integer	No	Truncate to N characters
Preserve Line Breaks	boolean	No	Keep original formatting

Outputs:

• text - Extracted text content
• wordCount, charCount - Text statistics
• truncated - Whether content was cut off

Extract Links

Extracts and filters all links from a webpage.

Inputs:

Field	Type	Required	Description
URL	string	Yes	The webpage URL
CSS Selector	string	No	Extract links from specific area
Filter Pattern	string	No	Regex to filter links
Include Internal	boolean	No	Include same-domain links
Include External	boolean	No	Include other-domain links
Maximum Links	integer	No	Limit returned links

Outputs:

• links - JSON array of link objects
• totalCount, internalCount, externalCount - Link counts
• link1 through link5 - First 5 links for easy access

Batch Fetch

Fetches multiple URLs in a single action.

Inputs:

Field	Type	Required	Description
URLs	text	Yes	URLs (newline or comma-separated)
Format	choice	No	Output format for all pages
CSS Selector	string	No	Applied to all pages
Timeout	integer	No	Timeout per page
Max Concurrent	integer	No	Parallel requests (1-10)

Outputs:

• results - JSON array of fetched content
• successCount, errorCount - Result counts
• result1_url, result1_content, etc. - First results

Triggers

Page Changed

Triggers when webpage content changes. Uses content hashing for reliable change detection.

Inputs:

Field	Type	Required	Description
URL to Monitor	string	Yes	The webpage to watch
CSS Selector	string	No	Monitor specific content
Ignore Whitespace	boolean	No	Skip formatting-only changes

Outputs:

• content - Current page content
• contentHash - MD5 hash for comparison
• checkedAt - Timestamp of check

Example Zaps

1. Save Articles to Notion

Trigger: RSS feed new item Action: Fetch Page (format: markdown) Action: Create Notion page with content

2. Monitor Competitor Pricing

Trigger: Page Changed (selector: ".price") Action: Send Slack notification Action: Log to Google Sheets

3. Build a Knowledge Base

Trigger: New Airtable record with URL Action: Fetch Page (format: som) Action: Update Airtable record with content

4. Content Aggregator

Trigger: Schedule (daily) Action: Batch Fetch (list of news sites) Action: Send digest email via Gmail

5. Broken Link Checker

Trigger: Schedule (weekly) Action: Extract Links from your site Action: For each link, Fetch Page Filter: Where fetch failed Action: Create issue in GitHub

Authentication

The integration supports two modes:

1. Plasmate Cloud - Enter your API key for hosted service
2. Local Plasmate - Leave API key blank, set base URL to your local instance

Rate Limits

• Plasmate Cloud: 100 requests/minute (default tier)
• Batch actions count as one request regardless of URL count
• Page Changed trigger polls every 5 minutes by default

Support

• Plasmate Documentation
• GitHub Issues
• Zapier Developer Docs

License

MIT