The situation: you wrote one good piece, then "repurposed" it by pasting the same 200 words into LinkedIn, X, and Instagram with the hashtags shuffled. Every feed can tell. The LinkedIn version wastes its "see more" fold, the X version has no standalone hook, and the Instagram caption buries the payoff three lines down.

The move: stop reformatting and start translating. Platforms have different registers, not different audiences for the same paragraph. Keep the brand voice constant and treat tone as the variable layered on top: LinkedIn is your voice with the collar buttoned, X is your voice after one coffee. Encode each platform's register as do/don't pairs lifted from real posts that performed, not adjectives, then have Claude produce a native post per platform from one source piece. The acceptance test: if you could swap two outputs between platforms and nobody would notice, the run flattened everything to the same neutral middle. Reject the batch and tighten the rules.

The Hub's Platform-Native Post Pack runs this as a single prompt: source piece in, per-platform set out. The design choice that makes it hold up is where the platform norms live: inline in the prompt as an editable rule block, not as a bet that Claude's training data is current. Platforms drift. When LinkedIn changes what the fold rewards, you edit one block and the whole pack updates. House rule that ships with it: re-check the norms quarterly, because this is the prompt that decays silently if you don't.

One rule to steal even if you never buy anything: add a refusal line. Tell Claude to refuse any hook built on "comment YES", "tag someone", or manufactured outrage about [YOUR TOPIC]. Engagement-bait gets reach today and costs trust tomorrow, and Claude will generate it without hesitation unless told otherwise.

Background reading on the thinking: Claude Skills for social media. The full Post Pack lives in the Hub, $149 one-time, own it forever.