LinkedIn Carousel Factory

Generate a complete 10-slide LinkedIn carousel JSON with hook-driven, concise content and .NET 8+ code examples for mid-to-senior .NET developers on a chosen...

Security Scan

The skill's stated purpose (generate LinkedIn carousel JSON) is coherent, but the runtime instructions ask the agent to write to local user paths and post to a Discord channel while declaring no credentials or installation requirements — a mismatch that could cause accidental data exposure or unexpected network activity.

Purpose & Capability

Name, description, and slide/code requirements all align: producing a 10-slide LinkedIn carousel with .NET 8+ code examples for mid/senior .NET devs. No extraneous libraries or binaries are requested, and generation-only behavior is appropriate for the stated purpose.

Instruction Scope

SKILL.md instructs the agent to write the generated JSON to a specific user path (~/Documents/Obsidian/ClawBrain/skills/draft/...) and to post a message to a Discord channel (#✅-tasks). These are explicit filesystem and external-network actions outside pure text-generation. The instructions also tell the agent to update a memory file (memory/what_works.md). The skill gives no details about how to authenticate or which service/integration will perform the Discord post, making the network/posting step ambiguous and potentially risky.

Install Mechanism

Instruction-only skill with no install spec and no code files; nothing is downloaded or written by an installer. This is low-risk from an install-supply-chain perspective.

Credentials

Declared requirements list no environment variables or credentials, yet the runtime instructions expect posting to Discord and writing to persistent memory. Posting to Discord or sending to collaborators would normally require a webhook, token, or an integration; the lack of declared creds is a mismatch. The skill also directs writes into the user's Documents/Obsidian vault and a memory file, which are reasonable for a content pipeline but should be explicit and consented to.

Persistence & Privilege

The skill requests writing generated drafts to a user Documents path and appending/recording data into memory/what_works.md (self-improvement notes). It does not set always: true nor request system-wide privileges. Writing to the user's vault and memory is persistent behaviour the user should be aware of, but it is not inherently privileged or covert.

What to consider before installing

This skill appears to do what it says (generate LinkedIn carousel JSON with .NET code), but review the following before enabling it: (1) It will write files into your ~/Documents/Obsidian/ClawBrain/... and update memory/what_works.md — confirm you want automatic writes to those locations and back up sensitive data. (2) It tells the agent to post to a Discord channel, but the skill declares no webhooks or tokens — determine which integration will perform that post and whether the agent has permission; if you don’t want network posting, remove or edit that step. (3) Because the skill generates code, inspect snippets for accidental disclosure or unsafe patterns before publishing. (4) If you want clearer authority boundaries, request the skill declare required webhooks/credentials or change the instructions to output only and require manual posting/review. If you want this cleared as benign, provide details about the Discord integration and confirm where the memory file is stored and who can read it.

Description

Generate a complete 10-slide LinkedIn carousel JSON for .NET developers. Takes a topic and produces a structured, hook-driven carousel ready for Claude Code ACP punch-up before publishing.

Use when

• OpenClaw trend scout has approved a weekly .NET carousel topic
• Lam requests a new carousel on a specific topic
• Engine 3 content pipeline triggers Sunday 9pm

Input format

Topic: [topic]
Audience: mid-to-senior .NET developers
Goal: [educate | provoke | warn | showcase]
Hook angle: [optional — if blank, agent picks strongest]

Output format

Return a JSON object with exactly 10 slides:

{
  "topic": "...",
  "hook": "...",
  "slides": [
    { "slide": 1, "type": "hook", "headline": "...", "body": "...", "code": null },
    { "slide": 2, "type": "problem", "headline": "...", "body": "...", "code": null },
    { "slide": 3, "type": "content", "headline": "...", "body": "...", "code": "..." },
    ...
    { "slide": 10, "type": "cta", "headline": "...", "body": "...", "code": null }
  ]
}

Slide types: hook, problem, content, comparison, warning, tip, cta

Slide rules

• Slide 1 (hook): Pattern interrupt. Must make a mid-senior .NET dev stop scrolling. Question, counterintuitive claim, or "here's what everyone gets wrong". Under 12 words.
• Slides 2–8 (content): Each slide = one idea. Headline under 8 words. Body under 40 words. Code snippets: .NET 8+, syntactically correct, no placeholder comments.
• Slide 9 (comparison or warning): Before/after or common mistake. Make the right way obvious.
• Slide 10 (CTA): Ask for one specific action — follow, comment with X, save for later. Not "like and subscribe."

Voice rules

• Audience: sceptical, experienced. Do not talk down to them.
• No AI phrases: "it's important to note", "leveraging", "game-changer", "delve", "in today's world"
• Concrete > abstract. Show the code, not the concept.

Write output to

~/Documents/Obsidian/ClawBrain/skills/draft/carousel-[topic-slug]-[date].json

After generation

Post to Discord #✅-tasks: "Carousel draft ready: [topic]. Review in vault → skills/draft/" Wait for Claude Code ACP punch-up before sending to Lam for approval.

Self-improvement instructions

After Lam approves a carousel: note what hook angle worked in memory/what_works.md under "Engine 3 — Content". After Lam revises: note what was changed and why.