borghei/Claude-Skills / tools/linkedin/linkedin-comment-writer
linkedin-comment-writer
VerifiedDrafts comments on other people's LinkedIn posts that add a specific the post lacked, then gates them offline. Use when commenting on a post, engaging with a prospect or peer, or checking a comment before pasting it.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 17.3 KB |
| assets/comment_brief_template.md | markdown | 2.5 KB |
| assets/sample_comment_drafts.json | config | 3.4 KB |
| assets/sample_story_bank.json | config | 2.1 KB |
| references/comment-moves.md | markdown | 10.6 KB |
| references/comment-quality-rubric.md | markdown | 8.7 KB |
| references/commenting-situations.md | markdown | 9.1 KB |
| scripts/comment_linter.py | script | 14.2 KB |
| scripts/comment_rules.py | script | 11.2 KB |
references/comment-moves.md
View on GitHub# Comment Moves
A move is the single job a comment does. Every comment worth posting does
exactly one. Two moves in one comment produce a short essay that nobody
finishes; zero moves produce applause.
The eight below are not templates to fill in. Each is defined by **what it
adds to the thread**, so the test for whether a draft is really that move is
whether the addition is present — not whether the sentences follow a pattern.
All people, companies, and figures in the examples are invented.
## How to choose
Start from what the user actually has, not from what would sound best.
| The user has… | Moves available |
|---------------|-----------------|
| Done the thing the post describes, with a result | Field report, Boundary, Price tag |
| Done it and got a different result | Counter-case, Boundary |
| Understanding of why it works, no data of their own | Mechanism, Open thread |
| Experience in a neighbouring field | Translation, Open thread |
| Nothing first-hand at all | Open thread — or skip |
| An answer to the question the post asks | Straight answer |
If two moves are available, prefer the one with the more concrete evidence.
A modest field report beats an elegant mechanism.
---
## 1. Field report
**Adds:** an outcome from someone who tried it.
**Use it when** the post recommends a practice and the user has run that
practice, or its opposite, long enough to see what happened.
**Shape.** Say what was done, at what scale, and what changed. Then say the
one thing that surprised the user. Stop there.
**Example.** The post argues that sales teams should publish pricing openly.
> We put prices on the site two years ago, for a product that sells between
> eight and forty thousand a year. Inbound volume fell by about a fifth and
> the share that turned into a real conversation roughly doubled. What I did
> not expect: the sales team's objection disappeared within a month, because
> they stopped doing the calls they hated.
**How it fails.** It turns into a case study with a beginning, middle, and
moral. Cut everything except what was done, what moved, and the surprise. It
also fails when the scale is missing — "it worked for us" from an unnamed
context tells the reader nothing about whether it applies to them.
---
## 2. Boundary
**Adds:** the condition under which the post's claim holds, and where it stops.
**Use it when** the post states a rule as universal and the user knows a
threshold — a company size, a price point, a team maturity, a market — past
which it flips.
**Shape.** Agree for the range where it is true, name the edge, say what
happens beyond it. This is the safest way to disagree with a prospect or a
senior person, because it concedes the main point.
**Example.** The post says daily stand-ups are a waste and should be deleted.
> True for us while the team was six people in one room. At fourteen across
> three time zones, the stand-up was the only place a blocked dependency
> surfaced before it cost a day. We kept it and capped it at ten minutes.
> I suspect the line sits wherever people stop overhearing each other.
**How it fails.** The edge is vague ("it depends on context"). A boundary
without a location is a hedge. Name the number, the stage, or the condition.
---
## 3. Mechanism
**Adds:** the cause behind a result the post only reports.
**Use it when** the author shows that something worked and either does not say
why or gives a reason the user thinks is incomplete.
**Shape.** Point at the result, offer the cause as a proposal rather than a
verdict, and say what would be true elsewhere if the proposal is right. That
last part is what makes it testable instead of clever.
**Example.** The post reports that shortening a trial from 30 days to 14
raised conversion.
> My guess at why: a 30-day trial lets the buyer put off the setup work until
> week four, by which point they have forgotten why they signed up. Fourteen
> days forces the import in the first few days. If that is the cause, you
> would see it in when people first load their own data, not in the last-day
> rush. Does the activation curve show that?
**How it fails.** It lectures the author about their own result. Keep the
verb tentative — "my guess", "I suspect" — and end by handing the question
back. It also fails when the proposed cause is unfalsifiable ("it creates
urgency"); if nothing would count as evidence against it, cut it.
---
## 4. Straight answer
**Adds:** an actual answer to the question the author asked.
**Use it when** the post closes on a question that is meant, rather than a
reflexive "what do you think?".
**Shape.** Answer in the first sentence. Give the one detail that shows the
answer is lived, not theoretical. Optionally, one line on what the user took
from it. Do not restate the question, and do not answer a different question
the user would have preferred.
**Example.** The post asks which hire the reader would make differently.
> Our first head of support. I hired for calm and empathy and should have
> hired for someone who would walk into engineering with a list. She was
> excellent with customers and the same forty tickets came back every month
> for a year, because nobody was turning them into fixes.
**How it fails.** The answer is abstract ("culture fit matters more than
skills"). The first sentence must be a concrete noun: a role, a tool, a date,
a decision.
---
## 5. Price tag
**Adds:** what the recommended approach costs, or what it breaks downstream.
**Use it when** the post presents a practice as all upside and the user has
paid for it — in money, time, morale, or a second-order effect.
**Shape.** Grant that it works. Name the cost, sized if possible. Say whether
the user would still do it. The last part matters: without it the comment
reads as a complaint.
**Example.** The post celebrates moving the whole company to a four-day week.
> We made the same change eighteen months ago and I would not go back. The
> part nobody mentioned beforehand: customer-facing teams cannot all take the
> same day, so we ended up with two rotas and a handover document every
> Thursday. That costs each of those people about an hour a week. Worth it,
> but it was not free.
**How it fails.** It tips into "well, actually". The move is a contribution
from someone who took the same decision, not a rebuttal — hence the explicit
"I would still do it" or "I would not".
---
## 6. Open thread
**Adds:** a question that only the author can answer and that the rest of the
thread would want answered.
**Use it when** the user has no first-hand evidence but can see what the post
leaves unresolved. It is the honest move when there is nothing to report.
**Shape.** One sentence showing the user understood the specific claim, then
one question about a detail the post omits: the sample, the exception, the
order of events, what happened next. The question must be answerable in a
sentence or two.
**Example.** The post says the author's company stopped doing performance
reviews and retention improved.
> Removing the annual review is the part everyone will quote. I am more
> interested in what replaced it for pay decisions. If managers now set
> raises without a written record, how do you check that two people doing the
> same job are not drifting apart over three years?
**How it fails.** The question is generic ("how did you measure success?"),
rhetorical (an objection wearing a question mark), or so broad it needs an
essay to answer. A good test: could the author reply from their phone in under
a minute, and would the reply contain new information?
---
## 7. Counter-case
**Adds:** evidence that the claim is wrong, or wrong more often than the post
admits.
**Use it when** the user has a specific case that contradicts the post and the
relationship can bear open disagreement (see the ceiling table in `SKILL.md`).
**Shape.** State what the user observed, in the same terms the post uses.
Say plainly that it conflicts. Offer the most likely reason the two
experiences differ. That third element turns a contradiction into something
the author can engage with instead of defend against.
**Example.** The post says cold email is finished as a channel.
> Our numbers disagree. Roughly a third of new customers last year started
> from a cold email, with reply rates that have been flat since 2023. The
> difference may be volume: we send about forty a week, each written by the
> person who will take the call. I would believe that the thousand-a-day
> version is finished.
**How it fails.** It attacks the author rather than the claim, it generalises
from one case as confidently as the post did, or it arrives with no evidence
at all. No case, no counter-case — fall back to Boundary or Open thread.
---
## 8. Translation
**Adds:** what the idea looks like in a context the author did not consider.
**Use it when** the post is sound within its field and the user works in a
neighbouring one where the same idea either transfers or breaks in an
interesting way.
**Shape.** The first sentence must name the author's claim, so the comment is
visibly about the post. Then describe the equivalent in the user's setting
and say whether it survives the move.
**Example.** The post, written by a software founder, says to ship before the
product feels ready.
> Shipping before it feels ready has an odd equivalent in clinical
> diagnostics, where I work: you cannot release early, but you can put the
> draft report format in front of three lab managers before a single assay is
> validated. We did that and threw away half the fields. The instinct
> transfers; the thing you ship just has to be the paperwork.
**How it fails.** It reads as "enough about you". If the first sentence does
not tie back to the author's claim, readers experience it as a subject change.
This is the least reliable of the eight for that reason — use it when the
parallel is close and the result is genuinely surprising.
---
## Rules that apply to every move
1. **One move per comment.** If a draft does two, split it and keep the one
with better evidence.
2. **The specific comes from the user.** A drafted figure the user has not
confirmed is a fabrication, however plausible.
3. **Open on the contribution.** No greeting, no compliment, no restating the
post. The author's first name at the start is optional and only when the
user actually knows them.
4. **Leave the product out.** Describe what the user does ("we run billing for
clinics"), never what they sell.
5. **End when the move is complete.** A closing question belongs only to the
Open thread and Mechanism moves, and there only when it is the point.
6. **Plain sentences.** No hashtags, no bullet lists, at most one emoji and
only if the user writes that way.
7. **Rotate.** In a session of several comments, no move twice running.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| slack-gif-creatoranthropics/skillsKnowledge and utilities for creating animated GIFs optimized for Slack. Provides constraints, validation tools, and animation concepts. Use when users request animated GIFs for Slack like "make me a GIF of X doing Y for Slack." | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |
| ponytailDietrichGebert/ponytailLazy senior dev mode for any coding task: the smallest change that fully solves it, and a reply a busy human understands in one read. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-debtDietrichGebert/ponytailHarvest every shortcut: comment into one debt ledger, so deferrals get tracked instead of forgotten. One-shot report. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-debtDietrichGebert/ponytailList every `shortcut:` comment (and older `ponytail:` ones) as a debt ledger. One-shot report, changes nothing. Use for "ponytail debt", "what did ponytail defer", "list the shortcuts", /ponytail-debt. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |