Unaware how slim the line between human and agentic coding is:

On my app doqment.dev I noticed LLMs typically start using search_content tool in hopes for direct content referrals, so I just improved it!

Now it actually returns useful information, with citations, and suggestions out of just natural sentence queries. 

Here an example: 

This makes up for a MUCH better agentic coding experience that I was expecting!

You can check it out at: doqment.dev/updates