- Home
- Skills
- Writing & Content
- LinkedIn Comment Drafts: Replies From Your Own Experience
More screenshots
Works with the AI Agents you already use
LinkedIn Comment Drafts: Replies From Your Own Experience
Paste LinkedIn posts or open a page yourself, get short comment drafts built from your own experience, ready for you to edit and post.
$5
LinkedIn Comment Drafts: Replies From Your Own Experience
Example session with this skill installed
Comment draft for this post, plus a connection note. Tom Okafor, Founder at Sockwise (sock subscription), 20h, 7 comments: "Honest question for subscription founders: how do you count churn when customers can pause? Every June our churn 'spikes' and every September it 'recovers'. I'm pretty sure half of it is people pausing for summer, but our dashboard lumps pause and cancel together. Do you count a pause as churn or not?"
- Read your context and instructions
- Compiled the linkedin comment drafts
LinkedIn comment drafts · 1 comment, 1 note · Voice: built-in style guard
1. Tom Okafor · Founder, Sockwise · 20h · 7 comments
Post: pasted text – counting churn when customers pause
Angle: answer his question from a real rebuild · Fact: F10 (coffee subscription churn report)
I'd count a pause as its own line in the churn report. When I rebuilt the churn report for a coffee subscription company, the old one counted paused subscriptions as cancelled, and separating them took monthly churn from 9.1% to 6.4% on the same data. Your June spike sounds like the same thing. Worth checking how many of the June "cancellations" are active again by September.
66 words · 378 chars · critic: APPROVE (round 2)
Connection note · Tom Okafor · send: tomorrow
Left a comment on your pause vs churn question. I build churn reports for small subscription brands and would like to stay in touch.
132/200 chars
Before you paste: read it once, change anything that isn't how you'd say it, paste by hand.
Facts used: F10, F21 (note: F01).
Connects securely to your tools. The creator never sees your data.
About this skill
LinkedIn Comment Drafts writes comments you'd actually want under your name. Paste the posts you're thinking of answering (text or screenshots) and it picks the 5-7 worth your time, then drafts one comment for each: 2-5 sentences, 40-90 words, opening with a concrete experience or a clear position. No "Great post".
It starts with a ten-minute interview that builds your fact bank: projects, mistakes, numbers with their source, opinions you hold. Each comment carries one detail from that bank, and the output lists which line it used. It won't invent a client, round a number up or claim work you didn't do.
It answers the author's question when there is one, ends on an action or a real question, and cuts the usual AI tics: "not X but Y" frames, aphorism endings, em-dashes, hashtags, emojis, links and pitches. One fact per comment, so a batch never repeats the same story, and an optional history file keeps you from commenting on the same author every day. Connection-request notes are optional, kept under the note limit (up to 200 characters on a free account, 300 on Premium; use the number your invite field shows), without a signature.
A separate critic reads every draft as the post's author and approves it or sends back exact fixes.
Draft-only. It never logs into LinkedIn, posts, comments, likes, follows, messages or sends invitations, and it doesn't open pages or run searches. By default you paste the posts. Optionally, if your agent has a browser and you've opened a LinkedIn page yourself, it can read up to about 20 posts from that page, read-only, after warning you that automated access is against LinkedIn's terms. You read each draft, change what you like and paste it yourself. No promise of reach, replies or leads.
If you have an Own Voice profile, it's used for tone. For a voice that learns from every edit you make, pair it with Own Voice Pro.
- Drafts only: you paste every comment and send every invite yourself. It never runs searches or opens pages.
- It doesn't fetch links. Paste the post text or a screenshot; a bare link gets a request for the text.
- Open-page mode needs a browser tool in your agent and reads at most about 20 posts from the one page you opened. Automated access is against LinkedIn's terms; this mode is at your own risk, and paste mode avoids it entirely.
- It stops reading at a login wall, captcha or unusual-activity warning.
- Comments are only as specific as your fact bank. With few facts, some posts get an opinion-only comment or are skipped.
- Post age and comment counts come from what you paste or the open page shows; if they're missing, selection uses the content only.
- No promise of reach, replies, followers or leads.
- Writes in the language of each post (English by default).
- Setup: a first-run interview of about ten minutes builds your fact bank; after that a batch takes a few minutes.
- Includes SKILL.md, references (selection, comment rules, cliche list, connection notes, interview), the critic prompt, a config template, a fictional example fact bank and a full worked example, all read by your agent as text.
- Uses an Own Voice or Own Voice Pro profile for tone if you have one.
- Two input modes: paste (default) or optional read-only open-page mode for a LinkedIn page you opened yourself.
- Single-user license.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 1 month ago
Permissions
File Scopes
Reads the posts you paste, your fact bank (~/.linkedin-comment-drafts/facts.md, or a folder you name) and optional config.md and history.md in the same folder. Writes facts.md there after the first-run interview, once you've reviewed it, and appends one line per comment (date, author, fact ID, topic) to history.md only if you say yes. If you have an Own Voice or Own Voice Pro profile in ~/.own-voice/, it reads it for tone and never changes it. Optionally reads your Freelance Bid Writer profile.md, only after asking, and never changes it. It never writes any part of the skill itself to disk. Pipes each draft into a one-line python3 (or wc) command to count words and characters, with no temporary file. In Claude Code it starts a fresh critic subagent and passes it the critic's instructions and your facts as text. In optional open-page mode, and only after warning you that automated access is against LinkedIn's terms, it reads the visible text of a LinkedIn page you opened yourself (up to about 20 posts, scrolling or expanding only to read them) and stores none of it. It makes no network calls of its own, never opens pages, runs searches or types into LinkedIn, never logs into LinkedIn, never posts, comments, likes, follows, messages or sends invitations, and never asks for passwords, cookies or logins.