LinkedIn AI skills

Bring: Your actual roles, projects, decisions, lessons and boundaries. Answer one question at a time; mark anything you cannot verify.

Keep in mind: It gathers material, not the final post. Missing facts stay empty. Keep the private Story Bank outside public repositories and exclude confidential client information.

Use linkedin-interviewer --mode bank. My role is [actual role] and my audience is [audience]. Ask one question at a time about real projects and lessons. Ask what may be shared publicly. Do not invent dates, numbers or outcomes. Leave unknown facts empty. Keep the Story Bank private; do not publish or commit it.

Bring: Paste your headline, About, Experience and Skills. Add screenshots or descriptions of your photo, banner and Featured section, plus your goal.

Keep in mind: It cannot fetch a profile URL, even with Apify. It can only assess sections you provide. Its recommendations are not verified conversion improvements or hiring predictions.

Use linkedin-profile-optimizer. Goal: [job search / clients / authority]. Audience: [audience]. Here are my pasted profile sections and visual descriptions: [content]. Use only confirmed facts. Do not fetch my profile URL or score unseen sections. Ask for missing evidence before proposing numbers or achievements.

Bring: Your audience, goal, themes, real Story Bank material, available posting days and time zone.

Keep in mind: A schedule is a proposal, not a forecast of reach. Adapt it to your capacity; the skill recommends a limited set of pillars and three to five posts per week.

Use linkedin-content-planner. Audience: [audience]. Goal: [goal]. Available days and time zone: [days, zone]. Confirmed material: [Story Bank excerpts safe to share]. Build a realistic seven-day plan. Flag facts still needed, explain timing assumptions and do not promise reach. Do not schedule or publish anything.

Bring: Team size, voluntary participants, business goal, current habits, brand boundaries and review responsibilities.

Keep in mind: It cannot ensure employee participation or business returns. Use consent, protect private work and avoid identical posts across employee accounts.

Use linkedin-employee-advocacy. Our team has [actual team size], with [voluntary participants]. Goal: [goal]. Review owner and boundaries: [details]. Create a fourteen-day pilot with individual voices, consent and realistic measurement. Do not invent ROI, require participation or publish on anyone's behalf.

Bring: Topic, audience, purpose, confirmed facts, preferred length and any examples of your voice.

Keep in mind: Hook formulas are writing aids, not viral guarantees. Review claims, voice and tone yourself. Publishing requires your approval and a separate configured publishing route, or manual posting.

Use linkedin-post-writer. Audience: [audience]. Topic: [topic]. Confirmed post spine: [Moment, Number if verified, Correction, Opposition, Ask]. Voice examples: [optional own writing]. Draft one useful post. Do not invent experience or outcomes. Ask for missing facts and show a draft only; I will review and post manually.

Bring: A draft for rewrite or audit, or three to six of your own posts for voice profiling.

Keep in mind: Post Audit is this skill's audit mode, not a separate skill. It cannot reliably evade AI detectors or establish authorship. You decide which edits preserve your meaning.

Ask the assistant to use linkedin-humanizer in audit mode. Review this draft without rewriting it: [draft]. Flag generic phrases, unsupported claims, readability and tone issues. Do not promise detector evasion. List questions about missing evidence. I will decide what to change and publish manually.

Bring: Pasted post text, or a post URL with an optional supported reading integration; your own topic for adaptation.

Keep in mind: A popular source post does not prove the hook caused its results. Do not copy the author's story, fabricate proof or reproduce engagement bait.

Use linkedin-hook-extractor on this pasted post: [text]. Explain the opening structure and give a blank template for my topic [topic]. Separate observed wording from assumptions about performance. Do not copy the author's personal story or claim the formula guarantees reach.

Bring: Source material you own or have permission to use, the intended audience and facts that must remain unchanged.

Keep in mind: It changes delivery, not facts or numbers. A transcript or pasted notes may be necessary when a source cannot be read. Credit others and preserve permissions.

Use linkedin-repurposer. This is my source material, or material I have permission to adapt: [content]. Audience: [audience]. Preserve these verified facts: [facts]. Rebuild it as a LinkedIn draft with attribution where needed. Ask if information is missing. Do not publish.

Bring: The post text, your actual perspective, your relationship to the topic and preferred tone. A URL needs supported reading access or pasted content.

Keep in mind: It cannot invent first-hand experience or infer the author's intent with certainty. A draft is not consent to post. Read the conversation before sending.

Use linkedin-comment-drafter. Post: [pasted text]. My actual perspective: [perspective]. Draft two thoughtful comments that add something specific. Do not claim experience I have not provided, insert a sales pitch or send anything. Ask for context if needed.

Bring: Post and comment text with authors and parent context, or supported URLs. Add your preferred tone and what you can truthfully answer.

Keep in mind: LinkedIn flattens replies to two levels. Missing parent context can change the meaning; verify the target and approved text before any external publishing.

Use linkedin-reply-handler. Post and parent context: [text]. Comments to review: [comments with authors]. My confirmed answer: [facts]. Draft relevant replies, identify which comment each addresses and flag unclear context. Do not send replies or infer approval.

Bring: Your recent comments, thread updates and timestamps, pasted manually or read through an optional integration.

Keep in mind: This runs when invoked or separately scheduled, not as an always-running daemon. Timing labels are workflow heuristics, not a proven optimal response window. No unattended replies.

Use linkedin-thread-monitor on these threads I participated in: [comments, updates, timestamps]. Identify author replies and sensible follow-up opportunities. Treat timing labels as heuristics. Do not run continuously, scrape unrelated accounts, schedule jobs or send anything. Ask for missing context.

Bring: A permitted engager list, or supported post URLs with optional Apify reading; your ideal customer profile criteria and known fields.

Keep in mind: These are audience-fit categories, not all profile visitors, hiring suitability or verified purchase intent. Missing company data remains unknown. Use consent and platform permissions, not intrusive mass outreach.

Use linkedin-engager-analytics. This list comes from posts I own or track with permission: [permitted data]. Audience-fit criteria: [roles, industries, companies]. Group peers, aspirational contacts, prospects and others. Mark unknown data and uncertainty. Do not infer profile visitors, hiring suitability or purchase intent. Do not contact anyone.