Doc Websites
Many documentation sites have little to no good search. Rather than crawling through endless pages on a poorly designed website - with Adrenaline you can add a single website URL and we'll scrape the entire site, index it, and answer your question.
Beyond the example sites, adding new sites is a premium-only feature.
The agent works by iteratively crawling through a documentation site. As it crawls the website, Adrenaline builds up a robust index of all the concepts and pages. By the time indexing is done, it can quickly retrieve an "understanding" of everything relevant to your question. The answer itself is then generated by GPT, which is fine-tuned to include key concepts from the docs.
Usage Tips
-
If you already know the document title, mention the specific document you're referencing!
- How do I implement the 'mergeSort' function? Search 'Sorting Algorithms'
-
Disambiguate vague concepts by referencing the section/subsection: Specify the section or subsection of the document you're referring to.
- How do I use the 'apply' function as mentioned in the 'Data Manipulation' section?
-
Mention the version of the library or framework: If the documentation is version-specific, mention the version you're using.
- How to use the 'map' function in version 3.0?
-
Include a direct quotation: If there's a specific sentence or paragraph you're unsure about, include it in your question.
- In the DEF library documentation, what does the phrase 'non-blocking asynchronous calls' mean in the 'Networking' section?