Metaphor Search
Access the internet's highest quality content. Recommended by people, powered by neural search.
Metaphor is a search engine plugin that retrieves high quality links for a given query. Specifically designed for queries that contain content recommendations, it returns both urls and titles for good examples of queries that follow this format. To ensure relevance, the plugin asks for at least 20 results for each query. Metaphor is a people-recommended and neural-powered search engine, ensuring the highest quality content. Note: queries should be written in singular form and the plugin requires a minimum of 20 results to operate effectively.
How to
Comments (0)
API docs
Learn how to use Metaphor effectively! Here are a few example prompts, tips, and the documentation of available commands.
Example prompts
-
Prompt 1: "Can you find me an article about the state of search?"
-
Prompt 2: "I want to learn how to draw. Can you recommend any resources?"
-
Prompt 3: "Can you provide a list of great Chinese restaurants in East Village?"
-
Prompt 4: "Who is Beethoven? I want to learn about him."
-
Prompt 5: "What are some of the most exciting modern artists?"
Features and commands
Feature/Command | Description |
---|---|
search | This command allows you to perform a search with a Metaphor prompt-engineered query and retrieve a list of relevant results. You need to provide the query string in the form of a declarative suggestion where a high-quality search result link would follow. Optionally, you can specify the number of search results to return, domains to include or exclude, and date filters. |
findSimilar | This command finds similar links to the provided URL. You need to provide the URL for which you would like to find similar links. Optionally, you can specify the number of search results to return, domains to include or exclude, and date filters. |
getContents | This command retrieves the contents of documents based on a list of document IDs obtained from either the /search or /findSimilar endpoints. You need to provide an array of document IDs. |