Skip to main content
Parallel.ai brings accurate web search, clean content extraction, site monitoring, and task runs into your workflows. It returns structured data with links and context you can pass to the next step, so you spend less time browsing and more time automating.
Use Parallel to search, extract, monitor, and run tasks on web data with reliable, automation-ready structured outputs.

​
How to Use MCP Nodes

​
What is Parallel MCP?

Parallel MCP creates a customized node that understands Parallel’s web tools, so you can use natural language to search the web, extract content, manage monitors, and run tasks. You describe what you want, the node handles the API details, and returns structured data that is easy to pass to the next step in your workflow.

​
What Can It Do for You?

  • Search the web and return ranked results with links and snippets for downstream processing
  • Extract clean article content from URLs with key fields like title and publish date
  • Monitor websites or queries for updates and retrieve event logs when changes are detected
  • Launch task runs and fetch their results as structured data

​
Available Tools

ToolWhat It DoesExample Use
ExtractExtract content from web URLs”Given page URL, extract the page and return title, publish date, and main text as structured data”
SearchSearch the web”Search the web for search query and return the top number of results results with title, url, and snippet as structured data”
List MonitorsList monitors”List up to max results monitors and return monitor id, name, target or query, and status as structured data”
Create MonitorCreate a web monitor”Create a web monitor named monitor name that tracks target query or URL at frequency check frequency, and return monitor id, name, and status as structured data”
Get MonitorRetrieve a monitor”Get the monitor with id monitor id and return id, name, target or query, and status as structured data”
List Monitor EventsList events for a monitor”For monitor id monitor id, list the most recent max events events and return event id, timestamp, and change summary as structured data”
Create Task RunCreate a task run”Create a task run for task description and return task run id, status, and created time as structured data”
Get Task RunRetrieve a task run”Given task run id, get the task run and return status, percent complete, and updated time as structured data”
Get Task Run ResultGet task run result (waits until complete)“Given task run id, wait for completion and return the final result as structured data”

​
How to Use

1

Create Your Parallel MCP Node

Go to your node library, search for Parallel, and click “Create a node with AI”
2

Add Your Prompt

Drag the Parallel MCP node to your canvas and add your prompt in the text box.
3

Test Your Node

Run the node to see the results. If it works as expected, you’re all set. If you run into issues, check the troubleshooting tips below.
4

Save and Reuse

Once your Parallel MCP node is working, save it to your library. You can now use this customized node in any workflow.

​
Example Prompts

Here are some prompts that work well with Parallel MCP: Web Search: 
Search the web for `search query` and return the top `number of results` results with title, url, and a 1 to 2 sentence snippet as structured data
URL Content Extraction: 
Given `page URL`, extract the page and return title, publish date, and main text as structured data
Create a Monitor: 
Create a web monitor named `monitor name` to track `target query or URL` every `check frequency`, and return monitor id, name, and status as structured data
List Monitors: 
List up to `max results` monitors and return monitor id and name as structured data
Monitor Events: 
For monitor id `monitor id`, list the most recent `max events` events and return event id, timestamp, and change summary as structured data
Task Runs: 
Create a task run for `task objective` and return task run id and status as structured data
Start simple and focused. Build one node that searches, another that extracts, and another that monitors. Passing structured data between smaller nodes keeps workflows fast, reliable, and easy to debug.

​
Troubleshooting

If your Parallel MCP node isn’t working as expected, try these best practices:

​
Keep Prompts Simple and Specific

  • Good: “Search the web for topic and return title and url for the top 3 results”
  • Bad: “Search for topic, open each result, extract key facts, create a monitor for updates, and summarize everything”
While this prompt might work, it is more efficient to break it into separate nodes. Parallel MCP works best with focused, single-action prompts.

​
Match What Parallel Can Do

  • Good: “Given page URL, extract and return title, publish date, and main text”
  • Bad: “Draft an email about topic and send it to recipient email”
Parallel MCP excels at high-accuracy web search, extraction, monitoring, and task runs. For sending emails, combine it with Gmail Sender node in your workflow.

​
Break Complex Tasks Into Steps

Instead of trying to do everything in one prompt (which can cause timeouts and errors): 
Search for sources on `research topic`, extract key facts from each `result URL`, then set a monitor to watch for updates and notify me
Break this into smaller, focused nodes that each handle one task:
1

Step 1: Search the Web

Search the web for research topic and return the top 5 results with title and url as structured data
2

Step 2: Extract Content

For each result URL, extract and return title, publish date, and main text as structured data
3

Step 3: Create a Monitor

Create a web monitor named monitor name to track research topic updates every check frequency, and return monitor id and status as structured data
In your workflow, connect these nodes sequentially. The result URLs output from Step 1 becomes the input for Step 2, and the monitor setup from Step 3 can reference the same topic or a specific URL from Step 2.

​
Focus on Data Retrieval

Parallel MCP is great at getting information from the web. For analysis or writing, connect it to other nodes. Example:
  • Good prompt: “Search for search query and return title and url of the top 5 results”
  • Bad prompt: “Search for search query, extract content, and write a 500 word summary”
Use the Ask AI node for summarization or analysis after Parallel returns data. This keeps each node focused and your workflow more reliable.

​
Troubleshooting Node Creation

If you’re seeing empty outputs in the node creation window (or if you’ve already created the node, hover over it and click “Edit”), use the chat interface to prompt the AI to add debug logs and verify the API response.
In the node creation window (or if you’ve already created the node, hover over it and click “Edit”), use the chat interface to describe in detail what you expected versus what you received.
First click “Fix with Gummie”. If multiple attempts do not resolve the issue, simplify your prompt or contact support.
MCP node creation often benefits from a few tweaks. Use the chat interface in the node creation window to refine filters, output fields, or pagination. The AI will adjust the node based on your feedback.

​
Need More Help?