A Claude Code content pipeline that turns one topic into a scheduled LinkedIn carousel.
The useful part is not “AI writes LinkedIn posts.” It is the end-to-end operating pattern: prompt → brand references → generated carousel slides → PDF → Cloudinary URL → Buffer scheduled LinkedIn document post.
What the video is actually showing
Input
A topic, often reused from an existing YouTube video title, plus brand/style reference images such as thumbnails.
Generation
Claude Code orchestrates image generation, slide planning, and LinkedIn post copy. The video claims use of OpenAI’s “GPT image 2” model.
Storage
Generated PNGs and the final PDF are stored locally and uploaded to Cloudinary for public URLs and searchable asset management.
Scheduling
Buffer is used instead of direct LinkedIn API posting because LinkedIn carousel/document posting is easier through a scheduler API.
Source limitation: this explainer is based on the available transcript. The creator says the GitHub repo is available through his free community; the repo itself was not inspected in this pass.
Pipeline map
content idea / YouTube title ├─ style-guide images │ └─ colors, fonts, thumbnail motifs, recurring layout language ├─ Claude Code worker │ ├─ researches/sets up Cloudinary API │ ├─ calls image model for each carousel slide │ ├─ stores local PNGs │ ├─ writes post copy + CTA │ ├─ stitches slides into carousel.pdf │ └─ calls Buffer API ├─ Cloudinary │ ├─ public PDF URL for scheduler │ ├─ searchable/generated asset library │ └─ optional edits/transforms later └─ Buffer └─ scheduled LinkedIn document/carousel post
The parts worth stealing
Reusable architecture
- Separate content generation from scheduling. Let Claude Code create assets locally, then hand off to stable APIs.
- Use a media store as the public URL bridge between local generation and third-party schedulers.
- Keep brand references in a folder so output style is repeatable across posts.
Agent workflow
- Start by asking Claude Code to set up one integration at a time: Cloudinary first, then generation, then Buffer.
- Make the agent explain where secrets should go instead of pasting API keys into chat.
- Add a specialized LinkedIn writing skill when generic post copy looks too AI-ish.
Operational caution
- The video is a demo, not proof of a production-quality content system.
- Direct posting/scheduling can create brand risk if copy, visuals, or CTAs are not reviewed.
- Cloudinary and Buffer credentials need explicit secret handling and redaction rules.
How this maps to Ananth / Dab
Daily Podcast / content repurposing
If Ananth creates transcripts, podcast notes, or YouTube/video material, this pattern can repurpose one source into LinkedIn carousel drafts, X threads, newsletter cards, or WhatsApp-friendly summaries.
Agent-managed publishing pipeline
Dab can orchestrate Claude Code workers that generate assets, but should keep final publishing gated: draft → preview → approval → schedule. That preserves autonomy without accidentally posting.
Potential workflow item
This is more than a read-it-later link. It is a candidate workflow pattern: “turn a captured source into platform-specific content assets with a review gate.”
Do not clone blindly
Only build it if there is a real publishing cadence. Otherwise, the valuable takeaway is the architecture: style references, asset store, scheduler API, and explicit human review.
Implementation checklist if Ananth wants this built later
- Define channel and cadence: LinkedIn only, or also X/Instagram/newsletter? Daily/weekly/manual?
- Create a brand kit folder: reference images, fonts/colors if known, examples of “good” and “bad” output.
- Pick storage: Cloudinary is sensible for generated assets and public URLs; configure secrets outside chat and commits.
- Pick scheduler: Buffer is the demonstrated path; compare against Typefully/Publer only if the publishing channel changes.
- Write generation specs: slide count, content structure, CTA style, forbidden visual patterns, and copy tone.
- Add review gate: generate preview HTML/PDF locally first; require Ananth approval before scheduling.
- Log artifacts: store generated previews and schedule results in a project/workflow folder so Dab can verify later.
Verdict
This is a strong seed for a Dab-assisted content repurposing workflow, especially if Ananth wants to turn longer source material into LinkedIn-ready carousel drafts. The immediate action should not be automatic posting; it should be a safe draft pipeline with Cloudinary/Buffer treated as optional execution adapters after the creative and approval loops are reliable.
Source URL: https://youtu.be/1q0RmehD8SU?si=JJU9Sar85PvWq-Kt