I think what should help you is defining more precise content plugs, plus using content: binaries to grab HTML file seems… Weird w.r.t naming. Link below should help more:
When it works in devmode, you can debug it easily using snappy-debug package which will point for you what needs to be adjusted.
