Works with the AI Agents you already use
Claude Code
Cursor
Codex CLI
Muse
Dots+22 moreClaude CodeCursorCodex CLIMuseDots
OpenClaw
Manus
Grok
Grok Bot+18 agentsThread Carousel
Automates the creation of Twitter-style Instagram carousels from research to PNG rendering.
Free
Thread Carousel
Example session with this skill installed
Make me a thread-style Instagram carousel about why most Reels lose viewers in the first 3 seconds. 8 slides, use my profile from config.json.
- Read your context and instructions
- Compiled the thread carousel
Opinion/method topic, so no research step: I wrote the copy directly. Profile loaded from config.json (name, @handle and avatar on every slide). 8 slides, all inside the 46px character budget, so the whole carousel renders at one font size.
The copy
| # | Job | Text (excerpt) | Visual |
|---|---|---|---|
| 1 | Hook | 🚨 Most Reels are lost before the first word is even said. |
And the problem isn't the algorithm. Here's what's happening 👉 | 2 screenshots side by side (yours) |
| 2 | What happens | People decide to keep watching in the time it takes to swipe, and that decision comes from the first frame, not from the caption | text only |
| 3 | The mechanism | → a frame that looks like every other video
→ a "hey guys" before the point
→ text that shows up after the second second | text only |
| 4 | Proof | your own retention graph showing the drop | 1 image (yours) |
| 5 | Why it matters to you | a strong idea with a weak opening reaches fewer people than a mediocre idea with a strong one | text only |
| 6 | De-escalation | But calm down, you don't need to become an editor. Moving the most interesting sentence to the start already changes the curve | text only |
| 7 | Zoom-out | this goes way beyond Reels: it's the same rule for a carousel cover and a YouTube thumbnail 😳 | text only |
| 8 | CTA | if you post short videos and want more of this, follow me | text only |
Checks the copy passed: no em dashes, max one CAPS word per slide, max 2 emoji per slide, arrows only on slides that set up the next one, CTA promises nothing the account doesn't already do.
Delivered
✅ Carousel ready
📁 Folder: carrosseis/reels-first-3-seconds/
🖼️ Slides: 8 files, 1080×1350 (size verified by the script)
👀 Preview: carrosseis/reels-first-3-seconds/preview.html
Images: 3 yours · 0 fetched · 0 generated on kie.ai
I opened the preview: no red TEXT DIDN'T FIT bar on any slide.
Next steps
- Look at
preview.htmland tell me any slide you want rewritten - Want a stock photo on slide 5? I can search Pexels (free) or propose a kie.ai prompt for your approval first
- Edit
spec.jsondirectly and re-run if you prefer to tweak the text yourself
Connects securely to your tools. The creator never sees your data.
About this skill
Generating high-quality Instagram carousels is a friction-heavy process involving copywriting, research, image sourcing, and layout adjustments. This skill automates the entire production of "Twitter-style" thread carousels, delivering ready-to-post 1080x1350 PNGs. It handles the specific constraints of the format, such as white backgrounds, short paragraphs, and visual proof blocks like news cards or citations.
The tool operates by strictly following a proven engagement structure. It researches factual topics via web search to ensure accuracy, generates specific copy designed for Instagram's reading rhythm, and manages visual assets. It uses a headless browser already present on your system to render the slides, avoiding heavy dependencies like Playwright or Puppeteer.
What it does
- Automated Copywriting creates structured slides with hooks, mechanisms, and custom CTAs.
- Visual Rendering produces 1080x1350 PNG files using a headless Chrome or Edge browser.
- Asset Management downloads news images, searches Pexels for stock photos, or generates conceptual art via API.
- Factual Verification performs web searches to validate numbers, dates, and sources before writing.
- Layout Enforcement includes news cards, citations with highlights, and profile headers.
How it works
- Define Topic by providing a theme or existing screenshots to the agent.
- Configure Profile by setting up a
config.jsonwith your handle, name, and avatar. - Approve Visuals by reviewing proposed prompts for generated images or selecting sourced photos.
- Render Carousel by running the build script which generates the PNG sequence and a
preview.html.
Frameworks & tools
Python, Headless Chrome/Edge, Pexels API, Kie.ai API.
Why this beats prompting it yourself
Standard LLM prompts often produce long, flowery text that breaks Instagram's layout and lacks the "human" rhythm of social threads. This skill enforces character budgets, prevents AI-typical phrasing, and provides a local rendering engine to guarantee the output is post-ready without manual design work.
Use cases
- Summarizing recent tech news or industry launches with verified sources.
- Creating educational "how-to" guides based on internal documentation.
- Turning opinion pieces or long-form threads into visual social content.
Known limitations
Requires a local installation of Chrome or Edge for rendering. Character counts are strictly capped to prevent text overflow, and font sizes are synchronized across all slides.
How to use
Skills on Agensi run inside your AI agent. Copy the prompt, paste it into your agent, and it connects to Agensi and installs this skill for you. If a payment is needed, your agent shows it to you first.
Reviews
1 people have installed this skill.
Trust & safety
Security scanned
Verified clean 12 days ago
- Free to download with an account
Requires
Kie
Pexels
Wikimedia
Jsdelivr
Node.js
Python