📄 The image above is the one-page quickstart — click it for the printable PDF.
A Claude Code / Claude skill that turns one product image into a complete, identity-consistent marketing funnel:
- a scroll-driven landing page,
- 3 Instagram Reels (9:16) with ready-to-paste captions, and
- a styled blog article
— all derived from a single approved Brand Brief, so every asset shares one visual and verbal identity instead of looking like three unrelated AI artifacts.
Note: the generated copy is German by default and the call-to-action is a demo placeholder (no real shop). Both are intentional defaults; change them in
SKILL.md(the "Locked defaults" section) if you want a different language or a real conversion goal.
image → short interview → Brand Brief → 🛑 you approve → builds everything
The skill is brief-first: it writes one Brand Brief (product identity, thesis line, OKLCH palette, fonts, tone, CTA, Reel angles) and shows it to you together with every asset prompt before spending any credits. You approve once, then it builds the landing page, the Reels, and the blog autonomously. The landing page is built first so its hero video can be reused (reframed to 9:16) for one of the Reels, saving generation cost.
| Requirement | Needed for | Notes |
|---|---|---|
| Higgsfield account + credits | All image/video generation | Sign up here: https://higgsfield.ai/s/mcp-arnold-oberleiter-ecZvus — an account with credits is required for any asset. |
| Higgsfield CLI | The generation engine | Installed automatically by the skill (curl … install.sh | sh). On Windows the installer needs a Bash shell (e.g. Git Bash). Then run higgsfield auth login once. |
| A browser + a static server | Landing-page QA & preview | The skill runs the page locally (e.g. python -m http.server) and verifies it. Any static server works; Python is just the default. |
| Internet access | CDNs / Google Fonts / Higgsfield | The generated pages load fonts and motion libs from CDNs. |
Everything else (the orchestration logic, the landing-page spec, the Reel/blog specs, and all templates) ships inside this skill. No extra files needed in your project.
This skill drives the Higgsfield CLI directly, so it works on its own. For the best asset quality it's recommended to also install the official Higgsfield skills — they add smarter model selection and the product-photoshoot prompt enhancer guidance:
higgsfield-generatehiggsfield-product-photoshoothiggsfield-soul-id(only if you want the same presenter face across all Reels)
Get them from the Higgsfield skills repo and drop them next to this skill.
Clone this repo straight into your skills directory as the skill folder:
# project-level
git clone https://github.com/Arnie936/image-to-campaign .agents/skills/image-to-campaign
# or user-level
git clone https://github.com/Arnie936/image-to-campaign ~/.claude/skills/image-to-campaignThen run higgsfield auth login once, and you're set.
Point the skill at an image and (optionally) a one-line idea:
"Mach mir aus diesem Bild eine komplette Kampagne:
./produkt.jpg"
It will ask at most two questions, write the Brand Brief, stop at the checkpoint for your approval, then build everything into a campaign folder:
kampagne-<slug>/
brief.md # the approved single source of truth
website/
index.html
assets/ # hero.mp4, middle.mp4, g1.jpg, g2.jpg, g3.jpg
reels/
reel-1-hook.mp4
reel-2-showcase.mp4
reel-3-lifestyle.mp4
captions.md # per reel: hook text, caption, hashtags, CTA
blog/
blog.html
assets/
README.md
image-to-campaign/
SKILL.md # orchestrator: bootstrap → interview → brief → checkpoint → build → QA → deliver
references/
brand-brief.md # how the brief is derived + the full asset-prompt list
landing-page.md # self-contained brief-aware landing-page build + browser QA
reels.md # 3 Reel angles, cinematic vs UGC, reframe reuse, captions format
blog.md # German SEO article spec + HTML templating
assets/
brief-template.md
blog-template.html
campaign-readme-template.md
README.md
Add your license of choice here before publishing (MIT is a common pick for skills).
