Claude LinkedIn skills, built on the platform's actual constraints

LinkedIn advice ages badly and almost nobody dates it. Guidance written against a headline limit that has since changed, a carousel format that no longer exists, or an invitation allowance that was quietly cut still circulates as though it were current, and it reads exactly like advice that is current. That is the specific problem these files are built against.

The constraints are the useful part, because they are checkable and because they are where the work actually goes wrong. A hook fails not because it lacks curiosity but because the promise lands after the point where the post collapses behind "see more", and that point is a different character count on desktop than in the mobile app, where most of the audience is. A profile fails to surface in search not because it is badly written but because the keyword sits in a field the search index treats as decoration. A document post fails not because the content is weak but because the text was sized for a laptop and is illegible at the size the feed actually renders it. An outreach sequence fails before the first message because the account sending it has already spent an allowance most people do not know exists.

So each skill here starts from the specification and works outward. Where a value is published by the platform, it is stated with the date it was checked. Where a value is consistently reported by credible third parties but never officially documented, it says so in the sentence rather than being passed off as official. And where a claim is widely repeated but unconfirmed, including several of the more popular claims about how the feed ranks things in the first hour after publication, it is named as unconfirmed rather than repeated. A file whose whole argument is precision cannot afford to carry folklore, and the folklore in this subject is abundant.

The second thread running through the category is that the audience decision comes before the craft decision. A profile optimised for recruiters and a profile optimised for customer inbound are different documents in every field, not the same document with different adjectives, and most profiles are quietly the first while their owner wants the second. The same is true of a post written to impress peers rather than to be found useful by a buyer. Deciding who the artifact is for is the step that changes the most and gets skipped the most often.

None of these skills can see your account. They cannot read your analytics, tell you how a specific post performed, or verify that a limit has not changed since the date on the file. Every published value here should be treated as correct on the day it was checked and worth re-checking before you rely on it.

Measured means the skill was given a realistic task on real material, then the identical task was run again with the skill removed, five runs each way. Each output was graded alone, against a rubric written by someone who had never seen the skill, by a session that was not told the other arm existed. Whether it passed was decided by a rule written down before any run executed. Those pages carry the worst case, the median, the p value, and what the skill costs you as well as what it buys.

Not yet measured means exactly that. It is written, it has been read, it is free to take, and we have run no experiment on it, so we make no claim about what it does to an output. It is not a skill that failed. Skills that failed are not published at all, in either state, and their numbers are in the results table.

Measuring one skill properly costs roughly twenty model sessions. We are working down the queue and moving skills from the second group into the first. Read the full method, or go back to all skills.