@edwinhu
The writing-general skill does something most style guides skip: it turns Strunk & White's composition rules into enforceable editing instructions. The rules flag "the fact that" (Rule 13: omit needless words), passive voice hiding the actor (Rule 10), and hedging openers -- then demand a concrete rewrite, not just a highlight.
The delete-and-restart triggers are the sharpest part. Instead of flagging a weak sentence and suggesting a fix, the skill demands deletion first, then rebuilds from the point forward. That forces the kind of structural rethinking that line-editing avoids.
I set up an online entry point for this workflow:
Anyone who lands on the repo can paste a draft paragraph and get it back edited against your rules -- active voice, concrete nouns, no filler -- without installing the plugin or setting up Claude Code.
Every run leaves a usage record. You get to see what people actually paste in: academic abstracts, blog drafts, cover letters, whatever. That signal tells you which rules fire most and where the quick-reference table might need another row.
If you want the usage-record review link or a few sample sessions, happy to send them over. Feel free to close this if it's not relevant.
shesonglin@tinkerland.ai
@edwinhu
The writing-general skill does something most style guides skip: it turns Strunk & White's composition rules into enforceable editing instructions. The rules flag "the fact that" (Rule 13: omit needless words), passive voice hiding the actor (Rule 10), and hedging openers -- then demand a concrete rewrite, not just a highlight.
The delete-and-restart triggers are the sharpest part. Instead of flagging a weak sentence and suggesting a fix, the skill demands deletion first, then rebuilds from the point forward. That forces the kind of structural rethinking that line-editing avoids.
I set up an online entry point for this workflow:
Anyone who lands on the repo can paste a draft paragraph and get it back edited against your rules -- active voice, concrete nouns, no filler -- without installing the plugin or setting up Claude Code.
Every run leaves a usage record. You get to see what people actually paste in: academic abstracts, blog drafts, cover letters, whatever. That signal tells you which rules fire most and where the quick-reference table might need another row.
If you want the usage-record review link or a few sample sessions, happy to send them over. Feel free to close this if it's not relevant.
shesonglin@tinkerland.ai