Love this — "anti-AI-slop" for design is exactly the right framing. AI output has a recognizable generic-ness, and encoding taste as explicit rules is the fix.
I built the writing-side equivalent for Chinese: stop-slop-zh. AI-generated Chinese has its own tells — filler phrases, punctuation habits, over-structured parallelism — and it's especially bad because most anti-slop work targets English. The skill bans specific phrases and structures, enforces a colloquial-phrase minimum, and runs a 4-layer QA pass.
hallmark does it for how things look, stop-slop-zh does it for how they read. Same underlying bet: AI doesn't lack capability, it lacks taste, and taste can be written down as rules.
Curious how you handled the tradeoff between hard rules (ban list) and softer judgment calls in hallmark — that's the line I kept wrestling with on the writing side. Great work on this.
Love this — "anti-AI-slop" for design is exactly the right framing. AI output has a recognizable generic-ness, and encoding taste as explicit rules is the fix.
I built the writing-side equivalent for Chinese: stop-slop-zh. AI-generated Chinese has its own tells — filler phrases, punctuation habits, over-structured parallelism — and it's especially bad because most anti-slop work targets English. The skill bans specific phrases and structures, enforces a colloquial-phrase minimum, and runs a 4-layer QA pass.
hallmark does it for how things look, stop-slop-zh does it for how they read. Same underlying bet: AI doesn't lack capability, it lacks taste, and taste can be written down as rules.
Curious how you handled the tradeoff between hard rules (ban list) and softer judgment calls in hallmark — that's the line I kept wrestling with on the writing side. Great work on this.