Free Claude skill

One idea in. A feed-ready infographic out.

This is the skill behind every graphic on my LinkedIn this summer. You give it one idea. It asks you 3 to 5 questions, agrees a layout with you, then builds the graphic and checks it against 15 rules before you see it. The animation on the left is the skill explaining itself.

Instant download. I will never spam you.

What is in the download

  • The process, as a SKILL.md Claude reads: brief, spec lock, wireframe, colour, gate, ship
  • A brand file and a reader file you fill in once, so every graphic comes out of the same system
  • The canon: design rules, technical gotchas, animation and export gotchas from around thirty shipped graphics
  • Seven scripts that measure the gate, capture frames, fetch real logos and export seamless GIFs
  • A working animated reference page and a starter set of real tool logos

It renders on your machine and writes local files. It never posts to LinkedIn and never invents a fact for a graphic.

Inside the zip

linkedin-infographic/         SKILL.md                  the process  brand.md                  your palette and type  references/                   canon.md                rules and gotchas    icp.md                  who it is for    animation-contract.md     examples/                     animated-reference.html   assets/logos/             8 real tool logos  scripts/                      audit.mjs               measures the gate    screenshot.mjs              capture-frames.mjs          fetch-logos.mjs             export-*.py                 setup.sh                installs what it needs

The six steps, in order

Most infographic prompts give you one finished image and a round of corrections. This one will not polish anything until you have approved the cheap decisions, and will not show you anything that failed its own checks.

  1. 1

    Brief

    It asks you 3 to 5 questions, then pulls the facts from your own material.

    Taste is easier to state up front than to correct afterwards. It never makes anything up.

  2. 2

    Spec lock

    You approve a title and a layout shape in one message.

    No pixels until you say yes. This is what stops the fourth round of "can we try it as a ladder".

  3. 3

    Greyscale wireframe

    You check the full layout in grey, with the real words in it.

    Moving a box now takes seconds. Moving it after the colour pass costs a rebuild.

  4. 4

    Colour pass

    Your colours, fonts and logos go on. They come from a brand file you fill in once.

    Every graphic comes out of one system instead of a fresh guess.

  5. 5

    The gate

    15 checks, pass or fail. 3 of them are measured by a script.

    One fail sends it back to step 4. You only ever see the version that passed, plus the table that proves it.

  6. 6

    Ship

    You get the PNG at 1080 x 1350, or a seamless looping GIF, plus the HTML it was built from.

    Change a word next month and re-export in seconds.

The gate, row by row

This is the table the skill fills in before it shows you a graphic. One line of evidence per row. One fail and you do not see that version. The 3 tagged rows are measured by a script.

#CriterionWhat passing means
1Dimensions and edgesmeasuredExactly the declared canvas, nothing clipped, 48px safe margin all round
2Header baked inTitle of at most 8 words plus a subtitle inside the image; it teaches without the caption
3Body text 15px or largermeasuredNo text below 15px at 1080-wide scale
4Hierarchy clearOne dominant focal point and an obvious reading order
5No redundant copyNo phrase repeated between header, cards and labels
6Unique visual per cardEvery card gets its own treatment; no row of identical boxes
7Connectors connectRails, arrows and flows unbroken and traceable in reading order
8Layout depicts the ideaA loop drawn as a loop, levels drawn as levels
9Real logosEvery named tool shows its real logo
10Palette disciplineYour palette only; no emoji, no gradient-heavy AI look
11No dead spacemeasuredNo stretched cards or filler; 90px of content in a 200px card is a fail
12Runnable by your readerThe reader you described could act on it; at most one deep technical panel
13House EnglishOne spelling convention; no em dashes in prose
14Premium finishAligned grid, consistent radii and shadows, subtle depth
15Value-forwardA cold reader learns something real from the image alone
A1Loop seamAnimated only. The first frame equals the last, measured against an adjacent-frame baseline
A2No opacity animationAnimated only. Everything visible at all times; motion is lift, scale, glow, colour, travel
A3Motion guidesAnimated only. Motion walks the reader through the content in order

The canon file in the download holds the rule behind each row. The skill adds to it every session, so it gets stricter with use.

★★★★★
Very few people can explain and teach AI in a way that is accessible, clear and fun. James is one of those rare few.
Tushar Agarwal
CEO at Hubble
★★★★★
Everything I needed: no filler, deeply interactive, and he clearly knows this space inside out.
Henrik Hagemann
Co-founder, Puraffinity
★★★★★
James had us building a personal AI coach in an hour. It opened my eyes to what is possible.
Michael Mangion
Founder

Ready when you are

Drop your email and the download unlocks right here, instantly.

Instant download. I will never spam you.

Questions people ask before running it

Does it work in Cowork, or only Claude Code?

Both. The steps, the brand file and the checks are plain instructions and run anywhere. The 3 measured checks and the GIF export use a small Node script that drives a headless browser. Claude Code installs it in one command. Cowork may not be able to, and the skill then judges those rows from a screenshot and tells you so. Static PNGs work in both.

What does it need installed?

A static PNG needs nothing beyond a browser. The measured checks need Node and Playwright, and the bundled setup script installs both. Looping GIFs also need Python with Pillow, and ffmpeg. The setup script lists anything missing and says what stops working without it.

Will the graphics look like yours?

No. Your palette, fonts and spelling live in a brand file you fill in on first run, and the reader you design for lives in a second file. Nothing of mine ships as a default. The examples on this page are what the same steps produced with my brand file.

Can it do animated posts?

Yes. Every frame is a function of time, so the loop joins without a jump and the export is identical every run. The download includes a working reference page. The animation at the top of this page came from it.

Does it write the LinkedIn post too?

No. It produces the image and hands it over with a one-line note of what it shows. The caption is yours.

Does anything leave my machine?

No. It renders locally and writes local files. The only network call is the optional logo fetcher, which visits the company homepages you list to pull their real logo.

Want the full playbook for turning a post like this into subscribers? The 3-step LinkedIn lead magnet funnel is the other half.

Want to build agents with me live?

The workshop showed you what's possible. The accelerator is where you actually build it. Eight chapters over four weeks, live sessions, office hours with your function peers, guest speakers, and a Demo Day at the end.

Week 1

Agent Foundations

Build your first agent and leave with it actually running. A Strategy Agent trained on your background, your goals, and how you think.

Security

Learn what can go wrong with agents before something does. Prompt injection, permissions, blast radius — and how to set sensible limits from the start.

Week 2

Integrations & MCP

Connect your agent to live data and the tools you already use. Your AI knows your pipeline now. It can act on it.

Skills

Package the things you keep doing into reusable workflows. Trigger them with a word. Share them with your team.

Week 3

Proactive Agents

Build agents that run while you're not watching. Scheduled briefings, overnight research, alerts that fire when something changes.

Team Agents

Get the rest of your team using this stuff. Share agents, share skills, share context. Stop being the only one who benefits.

Week 4

Coding Agents

Give an AI a brief and watch it write working software. No coding experience needed. Ship something real.

Demo Day

Show the cohort what you built. See what everyone else made. It's usually the best session of the four weeks.

🎤
Guest speakersFounders and operators from AI-native companies, every Thursday lunchtime
👥
Office hours by departmentWeekly peer sessions grouped by function: Founders, Product, Marketing, Sales, Ops
🏁
Demo DayEvery participant presents what they built at the end of Week 4
💬
The cohort100+ people building seriously, sharing what works and what doesn't

Join the Agent Accelerator. Four weeks, live with James and a cohort of founders and operators who are actually doing this.

Cohort 4 kicks off Tuesday 29 September 2026

Learn more about the Accelerator
Agentic Growth Framework