repeated marker patches

AI Chat Simulation

The server appends generated text by repeatedly streaming <template for="ai-tokens"> chunks. Each token patch includes a fresh <?marker name="ai-tokens"> so the next ordered token can append to the same range.

stream evidence


      

      

        
append marker

        
<template for="ai-tokens">
  <span>ready token</span>
  <?marker name="ai-tokens">
</template>

      

    

  


  

    
what to look for

    
The byte log records earlier tokens, then the selected bottleneck token, then dependent tokens. The iframe is the actual parser test: if the feature is supported, ordered token spans appear without a script concatenating strings into the chat bubble.

  


  

    
references

    
  


  

    
implementation reference

    
Need the exact API surface, compatibility boundaries, errors, lifecycle, and source links? Read the matching gendn reference ↗