Plasmate Integration for Make.com
April 12, 2026 ยท View on GitHub
A Make.com (formerly Integromat) custom app for Plasmate - the AI-optimized browser engine that provides 10-100x token compression for LLM processing.
Overview
Plasmate converts HTML to a Semantic Object Model (SOM) that dramatically reduces token usage when processing web content with AI. This Make.com integration brings Plasmate's capabilities to your automation workflows.
Key Benefits
- 10-100x Token Compression: Process web pages with significantly fewer API tokens
- AI-Optimized Output: SOM format preserves semantic structure while eliminating noise
- Multiple Formats: Choose SOM, plain text, or markdown output
- Batch Processing: Fetch multiple URLs in parallel
- Link Extraction: Extract and categorize links from any page
Installation
Prerequisites
Choose one of the following connection methods:
Option A: Plasmate Cloud API
- Sign up at plasmate.app
- Get your API key from the dashboard
Option B: Local CLI
- Install Plasmate CLI:
cargo install plasmate - Or download from releases
Installing in Make.com
- Go to your Make.com organization settings
- Navigate to Apps > Custom Apps
- Click Create a new app
- Upload the contents of the
app/folder - Add the modules from
modules/ - Configure your connection using
connections/
Manual Installation
- In Make.com, go to Organization > Custom Apps
- Click Create a new App
- Copy the contents of
app/base.jsoninto the app definition - Add each module by copying from
modules/*.json - Add connections by copying from
connections/*.json
Modules
Fetch Page
Fetch a web page and convert to AI-optimized format.
Inputs:
| Parameter | Type | Required | Description |
|---|---|---|---|
| URL | URL | Yes | The URL to fetch |
| Format | Select | No | Output format: SOM, text, or markdown (default: SOM) |
| CSS Selector | Text | No | Extract specific elements only |
| Timeout | Integer | No | Request timeout in seconds (default: 30) |
| Headers | Array | No | Custom HTTP headers |
Outputs:
| Field | Type | Description |
|---|---|---|
| content | Text | Processed page content |
| title | Text | Page title |
| description | Text | Meta description |
| links | Array | Links found on the page |
| tokenCount | Integer | Estimated token count |
| compressionRatio | Number | Compression ratio vs raw HTML |
Extract Text
Extract readable text content from a web page.
Inputs:
| Parameter | Type | Required | Description |
|---|---|---|---|
| URL | URL | Yes | The URL to extract text from |
| CSS Selector | Text | No | Extract from specific elements |
| Preserve Structure | Boolean | No | Maintain heading hierarchy |
| Exclude Selectors | Array | No | Elements to exclude |
| Max Length | Integer | No | Maximum character length |
Outputs:
| Field | Type | Description |
|---|---|---|
| text | Text | Extracted text content |
| wordCount | Integer | Word count |
| estimatedTokens | Integer | Estimated LLM token count |
| language | Text | Detected language |
Extract Links
Extract and categorize all links from a web page.
Inputs:
| Parameter | Type | Required | Description |
|---|---|---|---|
| URL | URL | Yes | The URL to extract links from |
| CSS Selector | Text | No | Extract from specific sections |
| Filter Pattern | Text | No | Regex pattern to filter URLs |
| Include External | Boolean | No | Include external links (default: true) |
| Include Internal | Boolean | No | Include internal links (default: true) |
| Max Links | Integer | No | Maximum links to return (default: 100) |
Outputs:
| Field | Type | Description |
|---|---|---|
| links | Array | Array of link objects (href, text, type) |
| totalLinks | Integer | Total links found |
| internalCount | Integer | Internal link count |
| externalCount | Integer | External link count |
Batch Fetch
Fetch multiple URLs in parallel.
Inputs:
| Parameter | Type | Required | Description |
|---|---|---|---|
| URLs | Array | Yes | List of URLs to fetch |
| Format | Select | No | Output format (default: SOM) |
| Concurrency | Integer | No | Parallel requests (1-20, default: 5) |
| Continue on Error | Boolean | No | Continue if URLs fail (default: true) |
Outputs:
| Field | Type | Description |
|---|---|---|
| results | Array | Array of fetch results |
| successCount | Integer | Successful fetches |
| failureCount | Integer | Failed fetches |
| totalTokens | Integer | Total token count |
Connections
Plasmate Cloud API
Best for most users. Serverless, no infrastructure required.
- Add the Plasmate Cloud API connection
- Enter your API key from plasmate.app
- Select your preferred region (or Auto)
Plasmate CLI (Local)
For self-hosted installations with maximum control.
- Add the Plasmate CLI connection
- Enter the path to your Plasmate executable
- Test the connection
Example Scenarios
Fetch and Summarize
Fetch a web page with Plasmate and send to OpenAI for summarization:
[Webhook] -> [Plasmate: Fetch Page] -> [OpenAI: Chat] -> [Google Sheets: Add Row]
See examples/fetch-and-summarize.json
Competitor Monitoring
Monitor competitor websites daily and notify on Slack:
[Schedule] -> [Plasmate: Batch Fetch] -> [Iterator] -> [OpenAI: Analyze] -> [Slack: Send]
See examples/competitor-monitoring.json
Link Crawler
Crawl a website and validate all links:
[Webhook] -> [Plasmate: Extract Links] -> [Iterator] -> [HTTP: HEAD] -> [Filter: Broken] -> [Email]
See examples/link-crawler.json
Token Savings Examples
| Source | Raw HTML Tokens | Plasmate SOM Tokens | Compression |
|---|---|---|---|
| News article | 15,000 | 800 | 19x |
| E-commerce product | 25,000 | 1,200 | 21x |
| Documentation page | 8,000 | 400 | 20x |
| Complex web app | 50,000 | 500 | 100x |
API Reference
Base URL
https://api.plasmate.app/v1
Authentication
All API requests require a Bearer token:
Authorization: Bearer YOUR_API_KEY
Rate Limits
| Tier | Requests/min | Concurrent | Monthly |
|---|---|---|---|
| Free | 10 | 5 | 1,000 |
| Pro | 100 | 20 | 50,000 |
| Enterprise | Unlimited | Unlimited | Unlimited |
Troubleshooting
Connection Failed
- Verify your API key is correct
- Check that the API endpoint is accessible
- For CLI connections, verify the executable path
Timeout Errors
- Increase the timeout parameter
- Some pages may be slow to load
- Consider using batch fetch with lower concurrency
Empty Content
- Check if the page requires JavaScript rendering
- Try a different CSS selector
- Verify the URL is accessible
Support
- Documentation: plasmate.app/docs
- GitHub Issues: github.com/nickshanks347/plasmate/issues
- Email: support@plasmate.app
License
MIT License - see LICENSE for details.