Writing Natural Language Queries


bloop is designed from the ground up to support natural language searches. By utilising semantic code search and Large Language Models like GPT-4, bloop allows you to ask questions like:

  • What does this repo do?
  • Is there a function that does X?
  • How does API endpoint Y work?
  • What happens if the state of the application is Z?

Natural language search is ideal for exploring unfamiliar codebases. It can locate code, summarize, explain, reason, and even suggest improvements.

Searching in Natural Language

To perform a natural language search, first ensure sure the natural language search type is selected in the dropdown menu to the left of the search bar in the header. Then, type your search query and press "Enter" to execute your search.

As semantic code search results load, a conversational dialogue will popout in a sidebar on the right hand side of the screen. Shortly after, a natural language response to your search will begin streaming.

NL Search

Continuing the conversation

To refine your query or make a related search, type your response in the textbox within the conversational dialogue.

To ask a new question or clear the dialogue, simply make a new search in the header.

NL Search


  • Frame your questions as if you were speaking to a colleague, rather than just entering keywords.

  • If at first you don't get a great result, try asking your question in a different way.

  • Refine your search by adding the lang and repo filters to the end of your search query, eg: lang:ts or repo:bloop.

  • As with most LLM-powered experiences, exercise caution with natural language search results. There is a possibility of encountering incorrect information or harmful content.

Previous: Getting StartedNext: Writing Regex Queries