higgsfield-studio-youtube-thumbnail
Skill1 starsChanged 3 days agoMITView on GitHub
Make high-click-through YouTube thumbnails and Shorts or Reels covers with Higgsfield, keeping faces and logos true. Triggers: 'make a YouTube thumbnail', 'thumbnail for this video', 'MrBeast-style cover', 'Shorts cover', 'Instagram video cover', 'change the face expression on my thumbnail'. Part of Higgsfield Studio: every render is priced first, never paid for twice, and written to a ledger. Needs Node 18+ and the higgsfield command line tool.
In this repository21 other files it gives its agents.
- audio-narrateSkill
- audio-transcribeSkill
- cloudflare-opsSkill
- dokploy-opsSkill
- elevenlabs-studioSkill
- hetzner-opsSkill
- higgsfield-studio-brandkitSkill
- higgsfield-studio-marketplace-cardsSkill
- higgsfield-studio-product-photoshootSkill
- higgsfield-studioSkill
- higgsfield-studio-soul-idSkill
- higgsfield-studio-video-explainerSkill
- higgsfield-studio-websitesSkill
- image-promptSkill
- namecheap-opsSkill
- plane-fleetSkill
- toolkit-credentialsSkill
- toolkit-updatesSkill
- video-promptSkill
- video-watchSkill
- zernio-opsSkill
The file
Show its rules and text
We split this skill into the rules it gives an agent
19 rules. About 0 are common, 9 are distinctive, and 10 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈9
Distinctive · estimate
This author's own choices. Worth a look.
10
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 37
Put --max N (a little above the price), --tag and --why "..." on every paid job.
- Very specific: line 38
Ask before spending more than about 10 credits, or 10% of what is left.
- Very specific: line 54
Create a clean thumbnail concept, generate each variant through the higgsfield CLI, inspect it, and make only requested surgical edits.
- Very specific: line 60
If higgsfield is missing, install it:
- Very specific: line 64
If higgsfield account status reports Session expired or Not authenticated, ask the user to run higgsfield auth login, then wait.
- Distinctive: line 75
Do not ask for facts already present in the brief.
- Distinctive: line 76
Never invent claims, outcomes, products, people, screenshots, or statistics that are not true of the video.
- Distinctive: line 77
Do not print raw JSON or job IDs to the user.
- Very specific: line 79
Do not use --count. Every concept, emotion, or camera take gets its own prompt and generation call.
- Distinctive: line 95
If the user gives an emotion count without names, use this ladder: shock, hype, rage, awe, laugh, fear, smug, charisma, confusion, …
- Very specific: line 99
Read references/thumbnail-frameworks.md.
- Distinctive: line 101
When a reference thumbnail exists, extract this structure before prompting:
- Very specific: line 126
Text: default No text, no readable UI labels, no watermark.
- Distinctive: line 128
Key elements: only signature props/effects that explain the information gap.
- Distinctive: line 130
Location: place, time, weather, and atmosphere when relevant.
- Very specific: line 165
Use Nano Banana Pro at explicit 4K.
- Distinctive: line 185
No stray text or watermark exists unless baked text was ordered.
- Distinctive: line 194
Use the picked completed job ID as the only image input.
- Very specific: line 205
If seedream_v5_pro is absent or rejects the submit, retry once with seedream_v4_5 --quality high.
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Higgsfield YouTube Thumbnail
A truthful concept first, then one render per variant, then small edits on the one the user picks.
What it costs
The main render is nano_banana_pro at 4K, quoted at 4 credits for 16:9 on 2026-10-12. A set of four variants is about 16 credits, and Higgsfield's guide allows up to 16 renders, which would be about 64. Offer one render first, or ask before a set. Small edits on the picked image use seedream_v5_pro at 2k (quoted at 2.5).
Before anything else
sh ../higgsfield-studio/scripts/setup-deps.sh --check
node ../higgsfield-studio/scripts/hf.mjs checkThe first checks Node and the higgsfield tool, and sh ../higgsfield-studio/scripts/setup-deps.sh installs what is missing after asking. The second checks the sign-in. If Higgsfield is not signed in, the user signs in once with higgsfield auth login, which opens the browser. Never print the sign-in token or open the CLI's credential file.
Spending credits
This skill is part of the Higgsfield Studio plugin. Every paid command runs through the main skill's guard, node ../higgsfield-studio/scripts/hf.mjs. It asks Higgsfield for the price before running, refuses above --max or below the reserve, never pays twice for the same request, and writes what each job really cost to .higgsfield/ledger.jsonl in the project. The full rules are in ../higgsfield-studio/SKILL.md. The short form:
- Price first with
--dry, using every parameter the real job will use. Tell the user the total for anything over about 5 credits. - Put
--max N(a little above the price),--tagand--why "..."on every paid job. - Ask before spending more than about 10 credits, or 10% of what is left.
node ../higgsfield-studio/scripts/hf.mjs accountshows the balance. - Anything outward or not undoable goes through
actand needs--yes. Say what will happen and wait for the user's yes before adding it. - Report in plain words, in the user's language: where the file is, what it cost, what is left. No JSON or job ids unless asked.
Free commands, such as lists, model get, presets and uploads, can run as node ../higgsfield-studio/scripts/hf.mjs run <command> or as higgsfield <command> directly.
Updates
This skill is versioned with its plugin. Its version is metadata.version in the header of this file. The plugin's history, for every skill in it, is in ../higgsfield-studio/CHANGELOG.md. To check for a newer version, open the address on its Latest: line and compare the top version with this one. Read every entry in between and tell the user what changed before anything is updated.
Higgsfield's guide
Everything below is Higgsfield's official guide for this task, with its paid and outward commands already pointed at node ../higgsfield-studio/scripts/hf.mjs. The rules above win where the two differ.
Higgsfield YouTube Thumbnail
Create a clean thumbnail concept, generate each variant through the higgsfield CLI, inspect it, and make only requested surgical edits.
Bootstrap
Before any generation:
- If
higgsfieldis missing, install it:sh ../higgsfield-studio/scripts/setup-deps.sh - If
higgsfield account statusreportsSession expiredorNot authenticated, ask the user to runhiggsfield auth login, then wait. - Confirm the locked model contracts when the catalog may have changed:
higgsfield model get nano_banana_pro --json higgsfield model get gpt_image_2 --json higgsfield model get seedream_v5_pro --json
UX rules
- Match the user's language. Keep CLI/model mechanics out of normal chat.
- Do not ask for facts already present in the brief. Ask one compact question only when a missing choice changes the result.
- Never invent claims, outcomes, products, people, screenshots, or statistics that are not true of the video.
- Do not print raw JSON or job IDs to the user. Deliver the image URL and a short variant label.
- A style-reference thumbnail is for visual analysis only. Never pass it with
--image; copying its identity or exact composition is forbidden. - Do not use
--count. Every concept, emotion, or camera take gets its own prompt and generation call. use_unlimis not a current CLI parameter. Never add--use-unlim; if the user explicitly asks to use an unlimited allowance, explain that this workflow must run on credits in CLI or through a surface that supports that allowance.
Intake gates
Collect only what the brief does not answer:
- The video's topic/title and the truthful promise the thumbnail may imply.
- Exact scene requirements, if any.
- Who appears: 0–3 people. If a concept needs a person and no face photo was provided, ask whether to use the user, another provided person, or a generic generated character. Never choose silently.
- Optional style-reference thumbnail. Analyze it with host vision for energy, framing, split layout, palette, and emotion; do not send it to Higgsfield.
- Optional logo and whether it stays flat or becomes a 3D object.
- Optional headline, 2–4 words. Default delivery is a clean image with no text. Use a deterministic overlay when the user requests an overlay; bake text into the generated image only when explicitly requested.
- Ratio:
16:9for YouTube by default,9:16for Shorts, or4:5for Instagram. - One final concept or a variant set. If unspecified and alternatives would materially help, offer a set of about four. Hard cap: 16 total generations.
If the user gives an emotion count without names, use this ladder: shock, hype, rage, awe, laugh, fear, smug, charisma, confusion, determination, disgust.
Concept gate
Read references/thumbnail-frameworks.md. Brainstorm at least five truthful concepts internally, across multiple frameworks, then select the strongest information gap with one focal subject and minimal clutter. Combine frameworks only when the result still reads in under one second at roughly 120px wide.
When a reference thumbnail exists, extract this structure before prompting:
brief, generic subject pose/action, elements, location, composition, background,
split (true/false), split_count, person_count, emotion, emotion_detailThe reference supplies art direction, never a specific identity. User instructions override it field by field.
Reference order
Pass face photos first in character order, then the logo. Repeat --image for every reference. When two or more references are attached, the prompt's first line must be a manifest such as:
IMAGE REFERENCES: image 1 = CHARACTER 1 face reference; image 2 = brand logo.Local paths are auto-uploaded. Previous completed job IDs also work as --image inputs.
Prompt contract
Assemble every main-render prompt in this order:
- Frame:
Bold, punchy YouTube-thumbnail composite — poster-grade, photoreal and high-impact, NOT a muted cinematic movie still, <ratio>, single unified frame — no split-screen, no diagonal divide, everything blends smoothly and organically across the same continuous shot.For9:16, addfaces in the upper two-thirds. A requested graphical representation replaces photoreal language with a clean diagram/graphic brief. - Scene brief: depict the user's exact content.
- Text: default
No text, no readable UI labels, no watermark.For explicit baked headline:TEXT: bold thumbnail headline text baked into the image, reading exactly "<TEXT>" — massive, ultra-legible sans-serif with a clean outline/glow treatment, placed where it never covers the subject's face. No other text, no watermark. - Subjects: large, foreground-dominant, chest-up or medium-close, filling about 40–60% of the frame. End with
All faces crisply sharp as the anchors of the shot. - Key elements: only signature props/effects that explain the information gap.
- Logo: preserve exact shapes, colors, proportions, and letterforms; keep it away from faces.
- Location: place, time, weather, and atmosphere when relevant.
- Composition: one power-third hero, clear scale hierarchy, depth, and strong subject/background separation.
- Background: vivid high-contrast color field or environment, soft vignette and edge falloff; do not divide it unless a split layout was explicitly requested.
- Lighting on people:
signature YouTube thumbnail lighting rig — strong key light sculpting the face, soft dreamy fill lifting shadows, and defined back light plus hair light tracing a clean bright rim around hair, shoulders and silhouette.Only the rim may use a colored accent. - Grade: vivid, bright, glossy, poster-punchy, deep blacks, crisp highlights, rich saturated colors, cohesive as one image. Restrain it only for an explicit calm/premium/muted brief.
For each photo-referenced person, include:
CHARACTER N: the person from attached face reference #K — IDENTITY LOCK: reproduce
this exact person with a photographic identity match — same bone structure, eye shape,
nose, lips, jawline, skin tone, hairline and hair texture. Do not beautify, average,
or restyle the face. Expression: <emotion phrase>.Split layouts
Use a split only when the user asks for split, before/after, versus, side by side, or the analyzed reference is split. A topical phrase such as X vs Y does not itself require a split. Replace the normal frame block with a clear halves/panels contract and keep all labels out unless short, truthful baked UI was explicitly requested.
Optional 3D logo
First create a 1:1 4K logo render, then use its completed job ID as the last --image on every thumbnail call:
node ../higgsfield-studio/scripts/hf.mjs create gpt_image_2 \
--prompt "Transform the attached 2D logo into a premium 3D logo render: extrude the exact logo shapes into glossy dimensional volumes; preserve every letterform, proportion and brand color; soft studio reflections, subtle bevels, crisp edges, clean dark neutral background, soft contact shadow, centered, generous margins, no extra text, no watermark." \
--image ./logo.png \
--aspect_ratio 1:1 \
--quality high \
--resolution 4k \
--wait --jsonMain render
Use Nano Banana Pro at explicit 4K. Write the final prompt to a temporary text file and pass it with --prompt-file so punctuation and multiline blocks are preserved safely:
node ../higgsfield-studio/scripts/hf.mjs create nano_banana_pro \
--aspect_ratio 16:9 \
--resolution 4k \
--image ./face-1.png \
--image ./logo.png \
--prompt-file thumbnail-prompt.txtOmit all --image flags when there are no references. For a variant set, make one call per distinct prompt. Keep the same references and settings; vary only the selected concept, expression, or camera-take line.
The completed JSON result contains id and result_url. Preserve both privately: the URL is delivered; the ID is the source for later edits.
Post-render gate
Inspect every result with host vision when available:
- Referenced identities visibly match.
- No stray text or watermark exists unless baked text was ordered.
- Explicit baked text matches character-for-character.
- The face/emotion and hero element remain readable at about 120px wide.
- The concept truthfully matches the video promise.
On a hard failure, retry the same prompt at most twice. If visual inspection is unavailable, do not claim it passed; deliver the result for user review. Present every passing variant and let the user pick before making optional tweaks.
Surgical tweaks
Use the picked completed job ID as the only image input. Keep the edit prompt narrowly scoped and state that every other pixel-level property remains unchanged.
node ../higgsfield-studio/scripts/hf.mjs create seedream_v5_pro \
--prompt "Change ONLY the person's facial expression to: <phrase>. Keep identity, face structure, hair, pose, body, clothing, logo, background, lighting and composition EXACTLY unchanged, pixel-faithful. Keep the YouTube thumbnail lighting rig intact." \
--image <picked_job_id> \
--aspect_ratio 16:9 \
--resolution 2k \
--wait --jsonIf seedream_v5_pro is absent or rejects the submit, retry once with seedream_v4_5 --quality high. For a 4:5 main render, Seedream has no 4:5; ask before changing the edit to 3:4, and disclose the crop/ratio change. Each accepted edit becomes the source ID for the next tweak.
CLI compatibility: versions through 1.1.20 can mislabel a nano_banana_pro job reference as nano_banana_pro_job. If the edit is rejected with a medias.0...data.type error, download the picked result_url to a local image and retry the same edit with --image ./picked-thumbnail.png. A local path is auto-uploaded as media_input; do not retry the invalid job-id payload.
Allowed tweak scopes: expression only, background replacement only, background recolor only, or rim-light recolor only. Never silently regenerate the full composition for a surgical request.
Text overlay
Keep the generated image text-free by default. When a headline overlay is requested, read references/text-overlay-bake.md and use one of its five presets: Beast, Fire, Neon Lime, Clean Glass, or Marker. The overlay path requires an environment capable of rendering HTML canvas; if unavailable, offer either the clean image or an explicitly approved baked-text regeneration. Never pretend an HTML preview is a flattened PNG.
Delivery
Return the passing result_url values with short semantic labels such as shock / close-up or product / size contrast. Mention the selected ratio and whether the deliverable is clean, overlay-ready, or text-baked. Do not expose internal prompts, job IDs, or retry mechanics unless the user asks.
Reference files
references/thumbnail-frameworks.md— 16 concept frameworks, information-gap rule, truthfulness law.references/text-overlay-bake.md— five deterministic text-overlay styles and 4K canvas-bake recipe.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

