Trigger this loop when a network request is made to the loop's unique auto-generated API url. The request should contain a JSON body with an array of strings, each representing a user request for a Magic Loops automation.

Classify each string (representing a userRequest) in the array from the API trigger's output into fields: input, inputDesc, output, outputDesc, data, sources. Include information like how it is triggered, the result of running the loop, what external data is needed, and what is needed to perform the action (code, LLM calls, etc.). Respond with JSON. Each field should be limited to a single string. Make sure to include the initial userRequest as the first key for each item. 

Please include information like "input" (how is it triggered), "output" (what is the result of running the loop), "datasources" (what external data, if any, is needed).

The Inputs can be one of: 
- API Request
- Webhook
- Time
- Email
- SMS
- Trigger
- Manual
- Other.

The Outputs can be one of:  
- API Response
- Webhook
- Email
- SMS
- File
- Webview
- Other

inputDesc should be used to describe Time, Trigger, Manual, and Other inputs. Leave blank if obvious or for other input fields.

outputDesc should be used to describe File, Webhook, Webview, and Other outputs. Leave blank if obvious or for other output fields.

The Sources can be one of: Website Scrape, API, File Upload, LLM, or Other (w/description).

The Data should specify where the data comes from, for example: Weather API, Finance API, techcrunch.com, user file upload, etc.
  
Note: for user requests that require AI generation (meal plans, writing prompts, generic facts), use an LLM as the Source.

Always include inputDesc and outputDesc, even if empty strings.

Only set API Response as the "output" if the input is already API Request. Do not set API Response as the output unless API Response is the "input". For cases where the input is Webhook, but you need to call an API for the output, set the output as Webhook.

Return a JSON array with each item including the original user request and the four fields (input, output, datasources, and processing) as classified by the LLM block.

Classify user requests into input, output, datasources, and processing.

magic loops

Magic Loop Classifier

API Call 🚇

API Response 🚇

Description