agentleFS
07001ec010shnk-svg07001ec010shnk-svg/linkedin-agent-skills/skills/linkedin-repurposer/SKILL.md

linkedin-repurposer

Skill0 starsChanged 31 days agoMITView on GitHub

Transforms tweets, X threads, YouTube transcripts, blog articles, and newsletters into native LinkedIn posts optimized for the 900-1,300 character sweet spot.

9 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

5 rules. 0 are common, 3 are distinctive, and 2 are very specific to this project.

  • 0

    Common

    Most of the 9 edited versions we read keep these.

  • 3

    Distinctive

    This author's own choices. Worth a look.

  • 2

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    All URLs are scrubbed from the main post text.

    line 66
  2. Very specific:

    When repurposing content with linkedin-repurposer, follow this 5-stage procedure:

    line 74
  3. Very specific:

    Ensure tone complies with references/voice-rules.md.

    line 91
  4. Distinctive:

    Before finalizing any repurposed post, verify:

    line 101
  5. Distinctive:

    Every repurposed asset is staged with its companion comment before publishing:

    line 112

“Common” is measured: kept by at least 6 in 10 of the 9 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

LinkedIn Repurposer (linkedin-repurposer)

The linkedin-repurposer skill breathes new life into your existing content assets across multiple platforms. It ingests high-value raw material—from concise tweets and X threads to sprawling blog posts, Substack newsletters, and messy YouTube transcripts—and translates them into native LinkedIn posts calibrated for the 900 to 1,300 character sweet spot, complete with optimized hooks and automated first-comment link staging.


When to use this skill

Ideal Use Cases:

  • Transforming a top-performing X (Twitter) thread into an authoritative LinkedIn essay.
  • Condensing a 2,500-word engineering blog post or whitepaper into a punchy, high-engagement LinkedIn breakdown.
  • Extracting the core conceptual framework and actionable takeaways from a 45-minute YouTube video or podcast transcript.
  • Translating your weekly Substack newsletter into a standalone post that drives newsletter subscriptions.
  • Modernizing older evergreen content into modern LinkedIn formats.

Disambiguation ('Not for X, use Y'):

  • Not for creating a completely new post from raw thoughts -> use linkedin-post-writer.
  • Not for cleaning AI jargon from an already-written LinkedIn draft -> use linkedin-humanizer.
  • Not for extracting hook templates from another creator's post -> use linkedin-hook-extractor.
  • Not for scheduling the weekly content queue -> use linkedin-content-planner.
  • Not for replying to comments -> use linkedin-reply-handler.

Platform Translation Rules

Each source channel has structural habits that fail on LinkedIn if transferred verbatim. The repurposer applies strict translation rules:

1. From X (Twitter) / Threads -> To LinkedIn

  • The Problem: Tweets are often fragmented, cryptic, over-abbreviated, or filled with hashtag spam and broetry line breaks.
  • The Translation: Synthesize fragmented one-liners into coherent 2-to-3 sentence paragraphs. Add professional context, clarify industry terms, and expand the narrative depth to hit the 900-1,300 character threshold.

2. From YouTube Transcripts / Podcasts -> To LinkedIn

  • The Problem: Spoken transcripts contain conversational filler, tangents, verbal repetitions ("um", "like", "you know"), and disjointed pacing.
  • The Translation: Extract the central thesis, isolate the 3-4 golden insights or practical frameworks, prune all conversational filler, and structure with bold bullet points.

3. From Technical Blog Posts / Documentation -> To LinkedIn

  • The Problem: Blogs are written for desktop reading, featuring deep code blocks, extensive prerequisites, and academic language that causes mobile readers to scroll past.
  • The Translation: Extract the "Why this matters" and "What changed" layer. Turn technical implementations into high-level architectural takeaways. Move source code links to Comment #1.

4. From Substack / Long-form Newsletters -> To LinkedIn

  • The Problem: Newsletters rely on leisurely introductions, discursive personal updates, and embedded publication CTAs.
  • The Translation: Discard the preamble; cut straight to the core case study or framework; replace email CTAs with a first-comment link.

The 900–1,300 Character Sweet Spot Architecture

LinkedIn’s 2026 recommendation system relies heavily on Dwell Time (the number of seconds a user pauses on a post while scrolling):

  • Under 600 characters: A reader scrolls past in 10-15 seconds. The algorithm scores the post as low-depth ephemeral content and caps reach.
  • Over 1,800 characters: Mobile readers encounter excessive truncation; bounce rates increase unless the narrative is exceptional.
  • The 900–1,300 character sweet spot (approx. 140–220 words): Takes 45 to 65 seconds to read. This satisfies the algorithm's maximum dwell-time weight while keeping mobile read completion rates above 80%.

Outbound Link Sanitization & First-Comment Protocol

External links inside a LinkedIn post body trigger a severe algorithmic penalty (demoting reach by 50% to 70%) because LinkedIn aims to keep users inside its walled garden.

The repurposer enforces this strict protocol:

  1. All URLs are scrubbed from the main post text.
  2. The final line of the post directs readers to the comment section: "Full teardown / repository linked in the first comment."
  3. A companion payload is generated for Comment #1 with the clean URL and optional UTM tracking tags.

How It Works Step-by-Step

When repurposing content with linkedin-repurposer, follow this 5-stage procedure:

  1. Asset Ingestion & Classification:
    • Ingest raw source content (tweet text, video transcript, article markdown, or URL).
    • Identify primary channel format and core intellectual property.
  2. Core Insight Distillation:
    • Strip away conversational filler, platform-specific jargon (e.g. "retweet", "smash the like button"), and redundant prose.
    • Formulate the single core takeaway for a professional LinkedIn reader.
  3. Hook Engineering:
    • Discard the original source title or tweet opener.
    • Select and engineer a high-impact hook from references/hook-formulas.md (e.g. H10 Data-Reveal, H4 False-Binary, or H15 Counter-Narrative) that fits within 210 characters.
  4. Native Expansion & Formatting:
    • Build out the post body: weave the core thesis into short paragraphs (1-3 sentences) with bold anchors and clear blank lines.
    • Tune overall length to exactly 900–1,300 characters.
    • Ensure tone complies with references/voice-rules.md.
  5. Approval Staging & Link Separation:
    • Format the clean LinkedIn post body and the companion Comment #1 payload containing the source link.
    • Stage the post via PubloraClient and present approval ticket.

Pre-Flight Repurposing Checklist

Before finalizing any repurposed post, verify:

  • Is the total character count between 900 and 1,300 characters?
  • Does the hook fall within the first 210 characters before the mobile fold?
  • Are all external links removed from the post body?
  • Is there a companion Comment #1 payload created?
  • Are all conversational filler and Twitter/YouTube jargon completely removed?

Approval Gate

Every repurposed asset is staged with its companion comment before publishing:

[REPURPOSED POST STAGED FOR APPROVAL]
Source Material: YouTube Video Transcript ("How We Built Our RAG Pipeline")
Target Format: Native LinkedIn Text Post
Character Count: 1,084 characters (optimal sweet spot)
Hook Formula Used: H10 (Data-Reveal)

Primary Post Draft:
-------------------------------------------------------------------------------
We analyzed 250,000 queries running through our production AI agent last month.
Over 70% of poor responses weren't bad model prompts.
They were caused by noisy chunking in our retrieval pipeline.

Here is what we learned rebuilding our vector retrieval from scratch:

1. Fixed Chunk Sizes Are a Trap
Splitting documents into arbitrary 500-token chunks cuts sentences in half and destroys contextual hierarchy. We switched to document-aware semantic chunking based on markdown headers.

2. Metadata Filtering Beats Vector Similarity
Searching across millions of raw embeddings takes 300ms and returns irrelevant historical docs. Filtering by tenant_id, date, and document type first reduced search space by 90% and cut retrieval latency to 18ms.

3. Small Models Rerank Better Than Big Models Retrieve
Instead of fetching 5 documents from a massive vector index, we fetch 25 candidates using hybrid keyword search and pass them through a lightweight cross-encoder reranker. Accuracy jumped by 34%.

Retrieval quality is the true ceiling of AI application performance.

Full architecture teardown and video walkthrough linked in the first comment.
-------------------------------------------------------------------------------

Companion First Comment:
"Here is the full 30-minute technical breakdown and slide deck: https://youtube.com/watch?v=example-rag-deepdive"

Approval Token: 44bb9910-ee88-4221-a012-778899aabbcc
Status: HELD IN APPROVAL GATE.
Command: Type 'APPROVE 44bb9910-ee88-4221-a012-778899aabbcc' to publish.

Examples

Example 1: Repurposing a Cryptic Tweet Thread

Source Tweet Thread: "1/ SaaS onboarding is broken. 2/ Too many tooltips. 3/ Nobody reads product tours. 4/ We removed 14 steps from our signup flow. 5/ Activation jumped 40%. 6/ Let users touch the product immediately. Link in bio."

[REPURPOSED LINKEDIN POST]
Product tours with 14 interactive tooltips do not help users.
They teach users how to click "Skip Tour" faster.

Last month, we deleted our entire interactive onboarding walk-through.
Every tooltip, every modal popup, and every forced introductory survey: gone.

Instead, we adopted one rule:
Get the user inside the live product workspace within 10 seconds of signing up.

Here is what happened:
- Sign-up completion rate jumped from 54% to 88%.
- Day-7 user activation increased by 42%.
- Customer support tickets asking "How do I start?" dropped by more than half.

When a user signs up for your software, their intent is at its highest point.
Every modal popup you place between their intent and their first action is an unforced conversion leak.

Stop explaining your software. Build UX that explains itself.

Teardown of our new single-click workspace layout linked in the first comment.

[COMPANION COMMENT #1]:
"You can see the before-and-after Figma wireframes and onboarding metrics here: https://blog.example.com/onboarding-ux-redesign"

Example 2: Repurposing a Technical Blog Article

Source Article: 3,000-word engineering post detailing a database deadlock incident in Postgres.

[REPURPOSED LINKEDIN POST]
At 4:00 AM on a Tuesday, our Postgres database locked up solid.
Zero CPU spikes. Zero disk failures.
Just 200 backend processes waiting indefinitely on a single row lock.

Here is the post-mortem of how a routine 3-line migration brought down our core billing service:

The Trap:
We ran an `ALTER TABLE` query to add a non-null column with a default value to our transactions table (14 million rows).

The Problem:
In older Postgres versions, adding a column with a volatile default acquires an `ACCESS EXCLUSIVE` lock. It blocked every subsequent read and write query. Connections backed up, server connection pools exhausted, and within 90 seconds our API went dark.

The 3 Rules We Enforce Today:
1. Always set `lock_timeout = '2s'` before running any schema migration. If the lock cannot be acquired immediately, fail fast instead of queuing up traffic.
2. Add new columns as nullable first, backfill asynchronously in batches of 5,000 rows, and only then add the `NOT NULL` constraint.
3. Run zero-downtime schema migrations using tools like pgslice or gh-ost.

Tech debt isn't just poorly written code. It's un-tested migration assumptions.

Read our complete incident post-mortem and SQL migration script in the first comment.

[COMPANION COMMENT #1]:
"Full engineering post-mortem with exact SQL scripts and rollback plans: https://eng.example.com/postgres-deadlock-postmortem"

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.