Higgsfield AI Skills
AI agent skills for image/video generation via Higgsfield AI, including Marketing Studio and Virality Predictor video scoring. Works with Claude Code, Cursor, Codex, and other AI coding agents that load Markdown-based skills.
Install
Pick one. Each method handles the Higgsfield CLI install and auth as part of skill setup.
npx skills — recommended, cross-agent
npx skills add higgsfield-ai/skills
Setup script
Universal fallback:
git clone --depth 1 https://github.com/higgsfield-ai/skills.git
cd skills
./setup
More options in INSTALL.md. Agent-driven install (paste into your agent): INSTALL_FOR_AGENTS.md.
Skills
| Skill |
Invoke |
Description |
higgsfield-generate |
/higgsfield:generate |
Image, video, 3D, and audio generation across 30+ models (Nano Banana 2, Soul V2, Veo 3.1, Kling 3.0, Seedance 2.0, Seed Audio 1.0, Flux 2, GPT Image 2, …), plus Marketing Studio for branded ads with avatars/products/hooks/settings and Virality Predictor scoring for finished videos. |
higgsfield-soul-id |
/higgsfield:soul-id |
Train a Soul Character — a reusable, face-faithful identity model. Returns a reference_id consumable by Soul-aware generation. |
higgsfield-product-photoshoot |
/higgsfield:product-photoshoot |
Brand-quality product imagery with mode-specific prompt enhancement. 10 modes (studio, lifestyle, Pinterest, hero banner, ad packs, virtual try-on, …) backed by gpt_image_2. |
higgsfield-marketplace-cards |
/higgsfield:marketplace-cards |
Marketplace product cards: compliant main image, secondary product images, and A+ style modules via backend prompt enhancement. |
higgsfield-websites |
/higgsfield:websites |
Build, edit, and deploy full-stack websites — React 19 + TanStack Start, server-rendered as one Cloudflare Worker with D1 / R2 / KV / Durable Objects / Containers. Create → get git repo access → edit locally → push → deploy preview/production, all via higgsfield website …. |
The skills chain: train Soul → use the reference id in generate (including Marketing Studio jobs). product-photoshoot and marketplace-cards are self-contained — backend enhances prompts before submitting image jobs. websites chains with generate to embed bespoke hero images, video loops, and OG assets in the site.
Modes
higgsfield-product-photoshoot — 10 modes for brand visuals:
| Mode |
What it's for |
product_shot |
Product on neutral / studio / catalog background |
lifestyle_scene |
Product in a real environment — hands, action, atmosphere |
closeup_product_with_person |
Tight crop with hands or partial face — beauty, demonstrating |
moodboard_pin |
Vertical 2:3 Pinterest-native pin, moodboard feel |
hero_banner |
Wide-format website / email / campaign header |
social_carousel |
3–10 connected slides for IG / LinkedIn / Facebook |
ad_creative_pack |
Coordinated pack of static ad variants for Meta / TikTok / Pinterest / Google |
virtual_model_tryout |
Product worn or used by an AI-rendered model |
conceptual_product |
Surreal / CGI-style / levitating / splash / sculptural product |
restyle |
Transform an existing image's aesthetic, mood, or seasonal context |
higgsfield-generate Marketing Studio — 9 modes for branded ad video:
| Mode |
What it's for |
ugc |
Default. Casual, organic-feel content from a presenter |
ugc_how_to |
Tutorial / explainer |
ugc_unboxing |
Unboxing reveal |
product_showcase |
Clean product highlight, polished |
product_review |
Presenter giving an opinion |
tv_spot |
Broadcast-style commercial |
wild_card |
Experimental, model picks the vibe |
ugc_virtual_try_on |
Trying on clothing — UGC vibe |
virtual_try_on |
Trying on clothing — polished, model-driven |
Quick Reference
| What you want |
Skill |
Note |
| Generate any image / video from a prompt |
higgsfield-generate |
Prefers gpt_image_2 / nano_banana_2 for images and seedance_2_0 for video by default |
| Generate audio from a prompt |
higgsfield-generate |
Prefers seed_audio by default |
| Image with my own face |
higgsfield-soul-id then higgsfield-generate |
One-time training, then --soul-id |
| Branded product photo (studio / lifestyle / Pinterest / hero / ad pack) |
higgsfield-product-photoshoot |
Mode-specific prompt enhancer + gpt_image_2 |
| Marketplace product cards / A+ style content |
higgsfield-marketplace-cards |
Main image, secondary images, and A+ style modules with hidden marketplace prompt templates |
| Branded ad video / UGC / unboxing / TV spot |
higgsfield-generate |
Marketing Studio mode with avatars + products + optional hooks/settings |
| Analyze a video's hook / attention / virality potential |
higgsfield-generate |
Uses Virality Predictor (brain_activity) with --video; returns score metrics plus an Open report link |
| Train a custom face identity |
higgsfield-soul-id |
5–20 photos, returns reference_id |
| Image-to-video animation |
higgsfield-generate |
Prefer seedance_2_0 with --start-image; use kling3_0 as lower-cost fallback |
| Build / edit / deploy a website, web app, landing page, or dashboard |
higgsfield-websites |
Full-stack React 19 + TanStack Start on Cloudflare; higgsfield website create/repo-access/deploy |
License
MIT — see LICENSE.
1---2name: higgsfield3description: AI agent skills for image and video generation via Higgsfield AI. Use this skill when asked to generate images or videos from a prompt, create branded ad video (UGC, unboxing, TV spot) with Marketing Studio, produce product photography or marketplace cards, train a reusable face identity (Soul Character), score a finished video's virality, or build/edit/deploy a full-stack website. Image/video generation spans 30+ models (Nano Banana 2, Soul V2, Veo 3.1, Kling 3.0, Seedance 2.0, Flux 2, GPT Image 2, …). Bundles five skills: higgsfield-generate, higgsfield-soul-id, higgsfield-product-photoshoot, higgsfield-marketplace-cards, and higgsfield-websites Markdown-based agents.4---5
6# Higgsfield AI Skills
7
8[](./LICENSE)
9[](./VERSION)
10[](#skills)
11[](https://discord.com/invite/higgsfield)
12
13AI agent skills for image/video generation via [Higgsfield AI](https://higgsfield.ai), including Marketing Studio and Virality Predictor video scoring. Works with Claude Code, Cursor, Codex, and other AI coding agents that load Markdown-based skills.
14
15## Install
16
17Pick one. Each method handles the Higgsfield CLI install and auth as part of skill setup.
18
19### `npx skills` — recommended, cross-agent
20
21```bash
22npx skills add higgsfield-ai/skills
23```
24
25### Setup script
26
27Universal fallback:
28
29```bash
30git clone --depth 1 https://github.com/higgsfield-ai/skills.git
31cd skills
32./setup
33```
34
35More options in [INSTALL.md](./INSTALL.md). Agent-driven install (paste into your agent): [INSTALL_FOR_AGENTS.md](./INSTALL_FOR_AGENTS.md).
36
37## Skills
38
39| Skill | Invoke | Description |
40|---|---|---|
41| [`higgsfield-generate`](./higgsfield-generate) | `/higgsfield:generate` | Image, video, 3D, and audio generation across 30+ models (Nano Banana 2, Soul V2, Veo 3.1, Kling 3.0, Seedance 2.0, Seed Audio 1.0, Flux 2, GPT Image 2, …), plus Marketing Studio for branded ads with avatars/products/hooks/settings and Virality Predictor scoring for finished videos. |
42| [`higgsfield-soul-id`](./higgsfield-soul-id) | `/higgsfield:soul-id` | Train a Soul Character — a reusable, face-faithful identity model. Returns a `reference_id` consumable by Soul-aware generation. |
43| [`higgsfield-product-photoshoot`](./higgsfield-product-photoshoot) | `/higgsfield:product-photoshoot` | Brand-quality product imagery with mode-specific prompt enhancement. 10 modes (studio, lifestyle, Pinterest, hero banner, ad packs, virtual try-on, …) backed by `gpt_image_2`. |
44| [`higgsfield-marketplace-cards`](./higgsfield-marketplace-cards) | `/higgsfield:marketplace-cards` | Marketplace product cards: compliant main image, secondary product images, and A+ style modules via backend prompt enhancement. |
45| [`higgsfield-websites`](./higgsfield-websites) | `/higgsfield:websites` | Build, edit, and deploy full-stack websites — React 19 + TanStack Start, server-rendered as one Cloudflare Worker with D1 / R2 / KV / Durable Objects / Containers. Create → get git repo access → edit locally → push → deploy preview/production, all via `higgsfield website …`. |
46
47The skills chain: train Soul → use the reference id in `generate` (including Marketing Studio jobs). `product-photoshoot` and `marketplace-cards` are self-contained — backend enhances prompts before submitting image jobs. `websites` chains with `generate` to embed bespoke hero images, video loops, and OG assets in the site.
48
49### Modes
50
51**`higgsfield-product-photoshoot`** — 10 modes for brand visuals:
52
53| Mode | What it's for |
54|---|---|
55| `product_shot` | Product on neutral / studio / catalog background |
56| `lifestyle_scene` | Product in a real environment — hands, action, atmosphere |
57| `closeup_product_with_person` | Tight crop with hands or partial face — beauty, demonstrating |
58| `moodboard_pin` | Vertical 2:3 Pinterest-native pin, moodboard feel |
59| `hero_banner` | Wide-format website / email / campaign header |
60| `social_carousel` | 3–10 connected slides for IG / LinkedIn / Facebook |
61| `ad_creative_pack` | Coordinated pack of static ad variants for Meta / TikTok / Pinterest / Google |
62| `virtual_model_tryout` | Product worn or used by an AI-rendered model |
63| `conceptual_product` | Surreal / CGI-style / levitating / splash / sculptural product |
64| `restyle` | Transform an existing image's aesthetic, mood, or seasonal context |
65
66**`higgsfield-generate` Marketing Studio** — 9 modes for branded ad video:
67
68| Mode | What it's for |
69|---|---|
70| `ugc` | Default. Casual, organic-feel content from a presenter |
71| `ugc_how_to` | Tutorial / explainer |
72| `ugc_unboxing` | Unboxing reveal |
73| `product_showcase` | Clean product highlight, polished |
74| `product_review` | Presenter giving an opinion |
75| `tv_spot` | Broadcast-style commercial |
76| `wild_card` | Experimental, model picks the vibe |
77| `ugc_virtual_try_on` | Trying on clothing — UGC vibe |
78| `virtual_try_on` | Trying on clothing — polished, model-driven |
79
80## Quick Reference
81
82| What you want | Skill | Note |
83|---|---|---|
84| Generate any image / video from a prompt | `higgsfield-generate` | Prefers `gpt_image_2` / `nano_banana_2` for images and `seedance_2_0` for video by default |
85| Generate audio from a prompt | `higgsfield-generate` | Prefers `seed_audio` by default |
86| Image with my own face | `higgsfield-soul-id` then `higgsfield-generate` | One-time training, then `--soul-id` |
87| Branded product photo (studio / lifestyle / Pinterest / hero / ad pack) | `higgsfield-product-photoshoot` | Mode-specific prompt enhancer + `gpt_image_2` |
88| Marketplace product cards / A+ style content | `higgsfield-marketplace-cards` | Main image, secondary images, and A+ style modules with hidden marketplace prompt templates |
89| Branded ad video / UGC / unboxing / TV spot | `higgsfield-generate` | Marketing Studio mode with avatars + products + optional hooks/settings |
90| Analyze a video's hook / attention / virality potential | `higgsfield-generate` | Uses Virality Predictor (`brain_activity`) with `--video`; returns score metrics plus an Open report link |
91| Train a custom face identity | `higgsfield-soul-id` | 5–20 photos, returns `reference_id` |
92| Image-to-video animation | `higgsfield-generate` | Prefer `seedance_2_0` with `--start-image`; use `kling3_0` as lower-cost fallback |
93| Build / edit / deploy a website, web app, landing page, or dashboard | `higgsfield-websites` | Full-stack React 19 + TanStack Start on Cloudflare; `higgsfield website create/repo-access/deploy` |
94
95## License
96
97MIT — see [LICENSE](./LICENSE).
starchild-ai-agent