LinkedIn Comment

A comment under a post with hundreds of reactions is seen by more people than most of the user's own posts, and a generic one costs credibility with the author.

Phase 1: Input

The user pastes the post text (plus the author's name and role if they have it), or a screenshot. A URL that cannot be opened: ask for the text. Never guess what the post said, and never scrape the feed with a browser. Read the Voice section of 01-Perfiles/Perfil Maestro.md.

Phase 2: Pick the type

Choose by what the post actually is. Never default to the one-liner.

#	Type	When	Shape
1	Add a datum	The post makes a claim you can back with a number	"We saw the same: 40% of our..."
2	The missing case	Right but incomplete	"This holds until {condition}. Then..."
3	Respectful disagree	You think it is wrong	The real agreement first, then the fork
4	Extend one line	One sentence is the good one	Quote it, build on it
5	The real question	The post skipped the hard part	One specific question
6	The receipt	You have done the thing	What happened, in two sentences
7	The correction	A factual error	Right, brief, kind and sure
8	The reframe	Right facts, wrong frame	"Another way to read this:"
9	The one-liner	Presence only	Under 12 words, funny or true

Phase 3: Write

• 2 to 4 sentences, one idea. Longer reads as a hijack, two points read as a blog attempt.
• Never open with "Great post", "Love this", "So true", "Couldn't agree more", "This resonates", "Gran post", "Excelente aporte", or the author's name with an exclamation mark. No emoji as the first character.
• Never restate the post. Say the specific thing: a comment that fits any post on the topic is noise.
• The data, receipts and experiences are the user's own. Ask when they are missing; never invent them.

Phase 4: Output

Two options of different types, labelled, with one line on which you would post. Run both through /content-humanizer (.claude/skills/content-humanizer/scripts/humanize.py, then detect.py from the same folder): in a short comment an em dash or a stock word stands out more than in a post.

COMMENT OPTIONS  (on @author's post about hiring)

[6 · Receipt]
We tried the no-resume version for 3 hires last year. Two were our best hires.
The third was a disaster, and the difference was whether they had done the job before.

[3 · Respectful disagree]
Agree on the signal problem. What worked for us was not dropping the resume
but giving every candidate the same paid 90-minute task.

Post the first: it is your own data and it concedes a failure, which is what people reply to.

Batch mode

For an engagement round, ask for the 5-10 posts as pasted text in one message and return one comment each in a single block. Keep a running list of who the user commented on this week in 01-Perfiles/LinkedIn Log.md (ask before writing): commenting on the same three people every day is visible.

Protocol

• Never post, like or comment for the user. The User Agreement (section 8.2) forbids bots and automated methods that comment on or like posts, and scraping.
• Credit: adapted from Jake Schincariol's linkedin-agent-skill (MIT).

Next steps

• /linkedin-post after the round: comments before posting help the post land.
• /personal-brand keeps the list of people worth engaging with.