メニュー
Replies to comments under the author's OWN LinkedIn post. Batch mode: processes multiple comments at once with phrasing variety. Critical for the algorithm — replying within the first 30 minutes = +64% comments.
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
SOC 職業分類に基づく
Checks whether an author's LinkedIn Article(s) are cited by LLM engines (ChatGPT, Perplexity, Google AI Overview, Claude.ai web). Runs a list of ICP queries, records citations, and computes share-of-voice. Run 4+ weeks after publishing an Article.
Builds a Voice Vocabulary for an author — a dictionary of their characteristic phrases, turns of speech, fillers, metaphors, and contrarian formulas. Used as a humanization layer on top of the DIP. Collected ONLY from raw sources (transcripts, voice input, personally written Telegram posts), NEVER from the author's LinkedIn posts.
Checks a LinkedIn post for factual accuracy. Finds statements compiled from multiple sources and verifies logical integrity. Use after creating a post or whenever facts need to be verified.
Checks content for generic / abstract phrasing, AI language, and repetition from the author's previous posts. Supports two-pass mode — the first pass catches original issues, the second (--second-pass) verifies that the humanizer didn't introduce new generic phrases.
Transparently shows where each key idea in a LinkedIn post comes from. Builds a source map. Lets you exclude specific sources.
Writes a LinkedIn comment on SOMEONE ELSE's post on behalf of an author. Loads the author's DIP, analyzes the post, picks a formula, and checks source grounding. Use when you need to comment on another person's post.
| name | reply-to-comments |
| description | Replies to comments under the author's OWN LinkedIn post. Batch mode: processes multiple comments at once with phrasing variety. Critical for the algorithm — replying within the first 30 minutes = +64% comments. |
Generate replies to comments under the author's own post. Batch mode with phrasing rotation.
$0 — author name (Seva, Kirill, Plurio)Author DIP:
Knowledge-Base/01-Identity-Profiles/[Author]-DIP.md
Comments-Engine:
Knowledge-Base/06-Comments-Engine/Comments-Engine.md
| Fact | Impact |
|---|---|
| Author replies within the first 30 min | +64% total comments, 2.3× views |
| Thoughtful reply (15+ words) | 2.5× algorithmic weight |
| Every reply = a new signal | The algorithm recomputes reach |
For each comment, identify the type:
| Type | Reply formula |
|---|---|
| Agreement / support | Expand + Add layer: "And what most people miss is [deeper insight]" |
| Question | Answer + Bridge: "Short answer: [X]. Real answer: [nuance]" |
| Disagreement | Acknowledge + Reframe: "I see your point, but [counter-evidence]" |
| Compliment | Redirect to insight: "The real credit goes to [specific detail]" |
| "How can I learn more?" | Give value + soft CTA: "Here's what I'd start with: [tip]" |
| Congratulations | From the micro-reply bank (rotate!) |
Rules for EVERY reply:
Rotation rule (CRITICAL):
Bank of micro-replies for congratulations:
Energetic:
Warm:
With character:
Minimal:
If teammates appear among the commenters, the reply should:
For each comment:
**Comment from [name]:** "[text]"
**Type:** [agreement / question / disagreement / compliment / congratulations]
**Reply:** [reply text]
**Formula:** [which formula was used]
At the end — a summary: