AI Social Media Content
NeverSight/learn-skills.dev
Create AI-powered social media content for TikTok, Instagram, YouTube, Twitter/X.
Generate AI avatar and digital persona video prompts for Seedance 2.0 on Higgsfield.
$ npx skills add rediumvex/ai-video-generator-claude --skill seedance-ai-avatar -a claude-codeProject install by default; add -g for ~/.claude/skills/.
$ gh skill install rediumvex/ai-video-generator-claude seedance-ai-avatar --agent claude-codeProject scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).
$ git clone --depth 1 https://github.com/rediumvex/ai-video-generator-claude.git skills-src && mkdir -p .claude/skills && cp -r skills-src/skills/09-ai-avatar .claude/skills/seedance-ai-avatar && rm -rf skills-srcUse ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.
Claude Code skills documentation · loads skills from .claude/skills/
Install the "seedance-ai-avatar" agent skill from https://github.com/rediumvex/ai-video-generator-claude/tree/main/skills/09-ai-avatar into .claude/skills/seedance-ai-avatar/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "seedance-ai-avatar", then confirm the skill loads.Claude Code copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$skill-installer install https://github.com/rediumvex/ai-video-generator-claude/tree/main/skills/09-ai-avatarType this inside Codex. $skill-installer <name> installs a curated skill from openai/skills. The installer writes to $CODEX_HOME/skills (default ~/.codex/skills). Restart Codex if the skill does not show up.
$ npx skills add rediumvex/ai-video-generator-claude --skill seedance-ai-avatar -a codexProject install goes to .agents/skills/; add -g for ~/.codex/skills/.
$ gh skill install rediumvex/ai-video-generator-claude seedance-ai-avatar --agent codexProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/rediumvex/ai-video-generator-claude.git skills-src && mkdir -p .agents/skills && cp -r skills-src/skills/09-ai-avatar .agents/skills/seedance-ai-avatar && rm -rf skills-srcUse ~/.agents/skills/ instead of .agents/skills for a personal install.
Codex skills documentation · loads skills from .agents/skills/
Install the "seedance-ai-avatar" agent skill from https://github.com/rediumvex/ai-video-generator-claude/tree/main/skills/09-ai-avatar into .agents/skills/seedance-ai-avatar/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "seedance-ai-avatar", then confirm the skill loads.Codex copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ npx skills add rediumvex/ai-video-generator-claude --skill seedance-ai-avatar -a cursorProject install goes to .agents/skills/; add -g for ~/.cursor/skills/.
$ gh skill install rediumvex/ai-video-generator-claude seedance-ai-avatar --agent cursorProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/rediumvex/ai-video-generator-claude.git skills-src && mkdir -p .cursor/skills && cp -r skills-src/skills/09-ai-avatar .cursor/skills/seedance-ai-avatar && rm -rf skills-srcUse ~/.cursor/skills/ instead of .cursor/skills for a personal install.
Cursor skills documentation · loads skills from .cursor/skills/, .agents/skills/, .claude/skills/, .codex/skills/
Install the "seedance-ai-avatar" agent skill from https://github.com/rediumvex/ai-video-generator-claude/tree/main/skills/09-ai-avatar into .cursor/skills/seedance-ai-avatar/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "seedance-ai-avatar", then confirm the skill loads.Cursor copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ gemini skills install https://github.com/rediumvex/ai-video-generator-claude.git --path skills/09-ai-avatar--scope user (default) or --scope workspace; --path is the subfolder of the repo that holds the skill; --consent skips the security confirmation prompt.
$ npx skills add rediumvex/ai-video-generator-claude --skill seedance-ai-avatar -a gemini-cliProject install goes to .agents/skills/; add -g for ~/.gemini/skills/.
$ gh skill install rediumvex/ai-video-generator-claude seedance-ai-avatar --agent gemini-cliProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/rediumvex/ai-video-generator-claude.git skills-src && mkdir -p .gemini/skills && cp -r skills-src/skills/09-ai-avatar .gemini/skills/seedance-ai-avatar && rm -rf skills-srcUse ~/.gemini/skills/ instead of .gemini/skills for a personal install, then run /skills reload.
Gemini CLI skills documentation · loads skills from .gemini/skills/, .agents/skills/
Install the "seedance-ai-avatar" agent skill from https://github.com/rediumvex/ai-video-generator-claude/tree/main/skills/09-ai-avatar into .gemini/skills/seedance-ai-avatar/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "seedance-ai-avatar", then confirm the skill loads.Gemini CLI copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ gh skill install rediumvex/ai-video-generator-claude seedance-ai-avatarInstalls for Copilot at project scope by default; add --scope user for a personal install. Preview a skill first with gh skill preview. Needs GitHub CLI 2.90.0 or later (public preview).
$ npx skills add rediumvex/ai-video-generator-claude --skill seedance-ai-avatar -a github-copilotProject install goes to .agents/skills/; add -g for ~/.copilot/skills/.
$ git clone --depth 1 https://github.com/rediumvex/ai-video-generator-claude.git skills-src && mkdir -p .github/skills && cp -r skills-src/skills/09-ai-avatar .github/skills/seedance-ai-avatar && rm -rf skills-srcUse ~/.copilot/skills/ instead of .github/skills for a personal install. Commit .github/skills so cloud agent and code review can use it.
GitHub Copilot skills documentation · loads skills from .github/skills/, .claude/skills/, .agents/skills/
Install the "seedance-ai-avatar" agent skill from https://github.com/rediumvex/ai-video-generator-claude/tree/main/skills/09-ai-avatar into .github/skills/seedance-ai-avatar/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "seedance-ai-avatar", then confirm the skill loads.GitHub Copilot copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ npx skills add rediumvex/ai-video-generator-claude --skill seedance-ai-avatar -a opencodeOpenCode documents no install command of its own. Project install goes to .agents/skills/; add -g for ~/.config/opencode/skills/.
$ gh skill install rediumvex/ai-video-generator-claude seedance-ai-avatar --agent opencodeProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/rediumvex/ai-video-generator-claude.git skills-src && mkdir -p .opencode/skills && cp -r skills-src/skills/09-ai-avatar .opencode/skills/seedance-ai-avatar && rm -rf skills-srcUse ~/.config/opencode/skills/ instead of .opencode/skills for a personal install.
OpenCode skills documentation · loads skills from .opencode/skills/, .claude/skills/, .agents/skills/
Install the "seedance-ai-avatar" agent skill from https://github.com/rediumvex/ai-video-generator-claude/tree/main/skills/09-ai-avatar into .opencode/skills/seedance-ai-avatar/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "seedance-ai-avatar", then confirm the skill loads.OpenCode copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
seedance-ai-avatarGenerate AI avatar and digital persona video prompts for Seedance 2.0 on Higgsfield.
Seedance AI Avatar is an agent skill from rediumvex/ai-video-generator-claude. Generate AI avatar and digital persona video prompts for Seedance 2.0 on Higgsfield. Use for virtual spokesperson content, digital twin videos, AI presenter clips, avatar-based marketing, virtual influencer content, or any video featuring a digital/AI-generated character as the main subject. Triggers on avatar, digital persona, virtual presenter, AI character, digital twin, virtual influencer, synthetic media, animated spokesperson.
Its SKILL.md is about 6.7k tokens, which your agent loads only when the skill is triggered. It is a single SKILL.md file with no bundled scripts.
It sits in Media & Creative, covering AI video generation and Influencer and creator marketing. It works with Seedance, Instagram and TikTok. The repository describes itself as: 10 Claude skills that generate studio-quality AI video prompts for Seedance 2.0 on Higgsfield. Viral hooks, SaaS demos, personal brand, faceless content, luxury aesthetic & more. The licence is MIT.
10 steps, taken from the step headings in SKILL.md.
Read from SKILL.md and the folder at commit ffdad7d. It shows what the files ask for, not the result of running them.
Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.
From allowed-tools in the SKILL.md frontmatter.
No scripts in the folder and no shell commands in SKILL.md.
From the folder's file list and the shell code blocks in SKILL.md.
No URLs in SKILL.md.
From URLs in SKILL.md, links to its own repository left out.
Names no API keys, tokens, secrets or passwords.
From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.
Seedance AI Avatar loads about 6.7k tokens when it runs. Until then it costs about 114 tokens; SKILL.md has 2,740 words of instructions outside code blocks.
Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.
The automated check found no risky patterns in SKILL.md.
Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.
The full file from rediumvex/ai-video-generator-claude at commit ffdad7d, republished under its MIT licence (© rediumvex). 2,740 words, ~6,692 tokens.
.claude/skills/seedance-ai-avatar/SKILL.md (or your agent's skills folder).A complete system for crafting Seedance 2.0 prompts on Higgsfield that feature AI avatars, digital personas, virtual presenters, and synthetic characters as the primary subject. This skill covers everything from photorealistic digital humans to stylized 3D characters, including hook patterns, environment design, camera work, lighting, and audio direction.
Before generating prompts, gather the following from the user:
Character definition
Content purpose
Technical requirements
Brand constraints
The uncanny valley problem — and how to avoid it
The uncanny valley is the psychological discomfort triggered when a synthetic human looks almost real but falls short in ways the brain cannot consciously articulate. Eyes that do not quite track. Skin that lacks micro-texture. Movement that is smooth but wrong. The brain pattern-matches to illness, deception, or death.
There are two reliable exits from the uncanny valley:
Exit 1: Go fully photorealistic. Push render quality, skin subsurface scattering, eye moisture, micro-expression, and hair fidelity past the threshold where the brain accepts the character as real. This requires explicit prompt language: "hyper-photorealistic skin texture," "subsurface light scattering on face," "wet corneal reflection," "micro-expressions," "natural breath movement."
Exit 2: Commit to stylization. A clearly stylized character — 3D cartoon, cel-shaded, geometric — never triggers uncanny valley because the brain never expected biological realism. Own the artifice. Make it beautiful and internally consistent.
The danger zone is the middle ground: a character that is clearly rendered but aspires to realism without achieving it. Avoid half-measures. Either push all the way to photorealism or lean fully into a defined non-realistic aesthetic.
Additional uncanny valley avoidance rules:
The most persuasive style for commercial and B2B content. Optimized for spokesperson videos, testimonials, and brand ambassadors.
Prompt markers: "photorealistic digital human," "hyper-realistic skin with visible pores," "subsurface scattering skin shader," "natural asymmetry in facial features," "wet corneal highlight," "professional studio grooming," "fine hair strands rendered individually"
Best for: SaaS product demos, financial services, healthcare, e-commerce spokesperson
Risks: Highest uncanny valley exposure — demand maximum fidelity or shift styles
Distinct personality with artistic intentionality. Strong brand identity potential. Avoids uncanny valley entirely.
Prompt markers: "stylized 3D character with smooth simplified geometry," "large expressive eyes," "exaggerated proportions with cartoon appeal," "cel-shaded render," "clean edge lighting," "Pixar-adjacent character design," "bold color palette skin tones"
Best for: Consumer brands, gaming, entertainment, apps, youth-oriented products
Risks: May feel low-trust for serious B2B content — calibrate style to audience
Flat or semi-flat illustrated character with motion. Highly distinctive, excellent for social media and explainer content.
Prompt markers: "2D animated character in motion graphics style," "flat illustration aesthetic," "bold outline strokes," "limited animation with expressive poses," "vector art quality," "anime-influenced design," "looping idle animation"
Best for: EdTech, productivity tools, startup explainers, social media content
Risks: Hard to convey facial nuance — lean on body language and motion design
The avatar is not a humanoid but a geometric entity — particle systems, fluid simulations, crystalline structures — that communicates through movement and light. Most experimental.
Prompt markers: "abstract digital entity formed from particles of light," "geometric crystalline structure with inner luminescence," "fluid simulation avatar morphing between states," "data visualization made sentient," "void-space entity with no fixed form"
Best for: Tech companies, AI products, concept content, high-concept brand films
Risks: Low relatability — pair with clear voice-over or text overlay
The first two seconds determine watch-through rate. For avatar content, there are four proven hook architectures.
Opens with visual corruption — static, scan lines, fragmented pixels, digital artifacts — then resolves into a clean, confident avatar. Communicates: synthetic origin, premium output.
Prompt language: "opens with heavy RGB chromatic aberration and scan line distortion, digital glitch artifacts fragmenting the face, resolves at 1.5 seconds into pristine clarity, sharp focus snapping to perfect render quality"
Use when: Launching an AI product, announcing new technology, establishing digital-native brand identity
The avatar materializes from nothing — particles assembling, wireframe filling, data streams converging into form. Communicates: creation, emergence, future.
Prompt language: "avatar assembles from scattered luminous particles converging to a central point, wireframe mesh fills with texture and color, form resolves from abstract to fully rendered in under two seconds, volumetric glow during formation"
Use when: Product launches, brand debut videos, origin story content, intro sequences
The avatar rapidly shifts visual style — sketch to 3D to photorealistic, or multiple artistic styles in sequence — before landing on a final defined look. Communicates: versatility, creativity, range.
Prompt language: "avatar transitions through four visual styles in rapid succession: pencil sketch, flat 2D illustration, stylized 3D, photorealistic render, each held for 0.4 seconds, final state held as primary look, transitions use dissolve with chromatic fringing"
Use when: Creative agencies, design platforms, AI tools, content creation products
Avatar makes direct, deliberate eye contact with the camera, creating the immediate impression of personal address. Can begin mid-gesture or mid-sentence — in media res. Communicates: directness, confidence, presence.
Prompt language: "avatar turns to face camera directly with deliberate intention, locks eye contact with subtle confident smile, slight lean forward, no introduction or preamble, as if resuming a conversation"
Use when: Sales content, coaching products, high-conversion landing page videos, personal brand content
The environment shapes perception of the avatar's authority, energy, and brand world.
The avatar exists in pure black, pure white, or a gradient void with subtle environmental effects (fog, particles, light beams). Maximum focus on the character. Minimum distraction.
Prompt language: "infinite dark void background, subtle volumetric fog at floor level, no horizon line, directional key light from upper left, ambient fill light from right, deep shadows behind avatar"
Best for: Premium product launches, minimalist brands, close-up spokesperson content
A clearly designed, branded environment suggesting a broadcast or content studio — but digital. Can include branded color washes, abstract architecture, floating graphic elements.
Prompt language: "virtual broadcast studio environment with clean geometric architecture, brand color accent lighting on background panels, subtle depth-of-field blur on background, professional lighting rig casting clean key light, polished floor with soft reflection"
Best for: B2B, SaaS, media companies, professional services
A recognizable workspace — desk, window, shelving — but elevated with futuristic design language. Grounds the avatar in a professional context while signaling technological advancement.
Prompt language: "near-future office environment with clean architectural lines, large floor-to-ceiling window with city view, ambient blue-white environmental light, holographic display elements visible in background, depth-of-field blur keeping avatar in sharp focus"
Best for: Enterprise, finance, leadership content, thought leadership
High-contrast urban digital environments — neon-lit corridors, data centers, glowing grid floors, cyberpunk architecture. High energy, high visual impact.
Prompt language: "neon-lit cyberpunk environment with deep shadows and saturated color pools of magenta and cyan, glowing geometric floor grid extending to horizon, electric particle effects drifting through air, haze atmosphere with volumetric light shafts"
Best for: Gaming, crypto/web3, youth brands, entertainment, tech startups
Camera work communicates status, intimacy, and energy for digital characters as much as for human subjects.
Framing fundamentals
Camera movement vocabulary
Slow push-in: builds anticipation and intimacy. "Camera slowly pushes in from medium to close-up over 4 seconds, no shake, smooth dolly motion"
Ken Burns on static frame: creates life from a still render. "Subtle slow zoom in with slight rightward drift, mimicking handheld documentary intimacy"
Orbit / arc move: reveals the avatar's three-dimensionality. "Camera arcs 30 degrees around avatar from left to right at shoulder height, smooth constant speed, avatar rotates slightly to maintain eye contact"
Reveal push: avatar is revealed as camera moves. "Camera starts on environment detail, pushes forward to reveal avatar in foreground, avatar turns to address camera at full reveal"
Match cut / style cut: avatar holds a pose that cuts perfectly to a different style or angle. Requires specifying the exact pose to hold at cut point.
Depth of field guidance
Stabilization language
The avatar emits or is surrounded by luminous energy — neon, bioluminescent, data-light. Strongly signals AI/digital nature.
Prompt language: "avatar lit from within by soft blue-white holographic glow, light emanating from chest area and face, ambient environmental light fills from all sides evenly, subtle rim light of electric cyan outlines the silhouette, no hard shadows"
Professional three-point lighting setup. Key light, fill light, rim light. The standard for commercial spokesperson content.
Prompt language: "professional three-point lighting: strong key light from upper left at 45 degrees, soft fill light from right reducing shadow contrast, tight rim light from behind right creating clean separation from background, color temperature 5600K daylight-balanced"
Hard, colored rim light from behind creates dramatic silhouette and strong graphic identity. Background colors bleed onto subject.
Prompt language: "neon rim lighting with magenta light from behind left and cyan from behind right, strong color separation, dark front face with colored halo around silhouette, moody underexposure on face revealing only key features"
Light behaves as physical substance — fog, haze, dust particles catching beams. Creates depth, atmosphere, and cinematic quality.
Prompt language: "volumetric light shafts cutting through atmospheric haze at 45-degree angle, dust and particle matter visible in light beams, soft falloff from bright to dark, environmental light wrapping avatar with soft fill from bounced volumetric scatter"
Lighting rules for photorealistic avatars specifically:
Sound design for AI avatar videos operates in three layers.
The avatar's voice is its most direct communication channel. Even in a video prompt (which describes the visual), the voice direction informs character and performance.
Voice direction markers to include in prompts or production notes:
Environmental sound that signals the digital world the avatar inhabits.
Textures to specify:
Background music that contextualizes the avatar's world and emotional register.
Music direction markers:
Sound design principle for AI avatars: sound should feel designed, not organic. Real-world ambience (birds, traffic, HVAC) undermines the digital world effect. Every sonic element should feel intentional and synthetic.
Photorealistic AI avatar of a professional woman in her early 30s, South Asian appearance,
shoulder-length dark hair, wearing a sharp minimal blazer in charcoal, positioned center-frame
in medium shot from chest up.
Avatar is placed in a virtual studio environment: clean architectural background with frosted
glass panels catching diffuse blue-white ambient light, polished light floor with subtle
reflection of subject, background held in soft bokeh blur.
Lighting: professional three-point setup — strong key light from upper left at 45 degrees,
warm fill from the right reducing shadow contrast, thin rim light from behind creating clean
silhouette separation from background. Color temperature 5600K.
Camera: static tripod shot at eye level, no movement, locked and authoritative. Shallow
depth of field with background in smooth bokeh.
Avatar's expression is open and confident — slight genuine smile, direct eye contact with
camera, a catch light visible in both eyes. Subtle asymmetry in the smile. A single strand
of hair rests slightly displaced to the right, adding naturalism.
At frame open she has just finished a thought and takes a brief natural breath before
beginning to speak. Head has a slight, natural tilt of attentiveness. No glitch effect —
clean professional entry.
Skin rendered with visible pore texture, subsurface light scattering at cheek and brow
areas, natural lip moisture highlight, eye corneal wetness with point source reflection.
Micromotion: almost imperceptible chest movement from breathing, barely visible blink at
second 2, slight weight shift that creates minimal shoulder movement. Everything reads as
alive, not frozen.
Color grade: clean and cool, slight desaturation, professional broadcast color treatment.Stylized 3D avatar character: gender-neutral presentation, large luminous eyes with
teal irises, smooth simplified skin geometry with clean subsurface warmth, short silver
hair rendered in individual strands, wearing a graphic oversized jacket in deep purple
with gold accent trim.
Opens with The Digital Birth hook: scattered luminous gold and teal particles
in void space, drifting inward as if being called together, density increasing,
wireframe mesh of the character's face emerging at center, particles filling in
surface texture, color and detail resolving over 1.8 seconds into the fully rendered
stylized avatar.
At full resolution the avatar opens its eyes — both irises catch light simultaneously —
and looks directly into the camera with a slow, deliberate smile. This moment is held
for 0.5 seconds in silence before any other motion.
Environment: pure void background, deep black with barely visible particle drift,
no horizon, rim lighting of teal from behind left and gold from behind right creating
dual-color silhouette framing that matches character palette.
Camera: medium close-up, centered, no movement during formation sequence, then very
slow push-in beginning at full resolution moment, arriving at close-up by second 5.
Lighting: rim glow from environment colors, soft fill from below and in front,
no hard shadows on face — clean, graphic, beautiful.
Music/sound: during particle formation, high-frequency shimmer building in intensity,
resolving to a single clean synth tone at full avatar reveal. Then minimal electronic
ambient pulse begins beneath subsequent content.
Cel-adjacent render with clean edge highlights, smooth gradient skin tones, bold
stylistic intentionality throughout. Never attempts photorealism.Photorealistic digital human avatar, male, mid-40s, Northwestern European appearance,
close-cropped dark hair with salt-and-pepper at temples, wearing a clean white shirt
under a structured dark navy jacket, no tie. Commands immediate authority.
Opens in media res with The Fourth Wall Break: avatar is already mid-turn toward camera,
as if he became aware of our presence. He completes the turn in the first 0.8 seconds,
makes direct, unhurried eye contact, and leans forward almost imperceptibly — the body
language of a person with something important to say, not a presentation to deliver.
Environment: near-future corporate interior, large window occupying the entire background,
out-of-focus cityscape in deep-focus bokeh, the room's architecture is clean and minimal —
single floating desk edge visible at frame right, ambient indirect ceiling light from above,
blue-hour exterior light bleeding through window glass and catching the side of his face.
Lighting: ambient daylight from window provides cool atmospheric fill, key light from
left at 45 degrees slightly warmer to balance the cool ambient, crisp but narrow rim
light from the window side. High dynamic range — bright window, controlled shadows on
face, catch light clearly visible in both eyes.
Camera: low eye-level camera — slightly below eyeline, reinforcing the subject's authority.
After initial static hold (0–3 seconds), a nearly imperceptible slow push-in begins,
arriving 8% closer by second 10. This movement is felt subconsciously, not noticed.
Skin: maximum photorealistic fidelity — visible pore texture, subsurface warmth at
cheeks and forehead, five o'clock shadow rendered at individual hair level, natural
lip texture, oil/moisture on brow area.
Micromotion: regular natural blinks every 3–4 seconds, subtle jaw movement as if
processing a thought before speaking, minimal but present chest breath movement,
a very slight asymmetric quality to his resting expression — the face of someone
thinking, not posing.
Sound: room ambience is complete silence except for a very low server-hum that suggests
he exists at the intersection of physical and digital. No music until he begins speaking.
Color grade: slightly desaturated with lifted blacks, clean digital cinema aesthetic,
neutral whites, no stylization that would undermine the photorealistic character intent.Rule 1: Commit to a style lane and stay in it. Mixed signals between photorealism and stylization produce uncanny valley. Every detail in the prompt must serve the same aesthetic target. If you write "photorealistic skin" and "cartoon eyes" in the same prompt, the output will be incoherent.
Rule 2: Describe aliveness, not perfection. Perfect symmetry, frozen expressions, and stillness read as dead or rendered. Include asymmetry, micro-motion, imperfection, and breath. "A strand of hair slightly displaced" does more for believability than five sentences about render quality.
Rule 3: Eyes anchor everything. The eyes are the first thing viewers evaluate for believability and connection. Always specify: catch light presence, iris color and texture detail, corneal wetness, natural blink timing, and where the gaze is directed. Underspecified eyes produce the single most common uncanny valley failure.
Rule 4: Ground the avatar to its environment. The avatar should interact with its environment through light and shadow — light from a window should fall on the skin, a neon environment should tint the avatar's face. A character that floats in front of its background without environmental interaction reads as a composite, not a presence.
Rule 5: Camera language is character language. How the camera frames and moves around an avatar communicates personality and status before the character does anything. A locked-off shot says authority. A slow push-in says intimacy. A low angle says power. A slight handheld wobble says authenticity. Choose intentionally.
Rule 6: Use sound as world-building, not decoration. Synthetic ambience and designed sound textures establish that this is a digital world, which paradoxically makes the avatar more believable — you are setting context for synthetic existence rather than trying to pass it off as real.
Rule 7: Specify the hook in the first sentence of the prompt. Seedance and most AI video models weight the opening of the prompt heavily for the opening of the video. Put the hook description first — what happens in seconds 0–2 — before any other detail.
Rule 8: Duration shapes detail density. For 2–4 second clips: focus entirely on one moment, one expression, one action. Every detail must earn its place. For 6–10 second clips: allow for a single transition or beat — hook, then hold. For 15–30 second clips: you can describe an arc — opening state, middle development, closing beat.
Rule 9: Avoid "AI-looking" descriptions. Paradoxically, describing an avatar as "AI-generated" or "digital" in the prompt often produces visual clichés — blue grids, floating data, lens flares on eyes. If the goal is photorealism, describe a real human. If the goal is stylized, describe a designed character. Let the aesthetic communicate the AI identity, not the text.
Rule 10: Test photorealism with the close-up rule. If you are writing a photorealistic avatar prompt, ask: does this description hold up in extreme close-up? If the answer is no — if the pores, eye detail, hair texture, and micro-expression are underspecified — push harder on those elements before finalizing the prompt.
© rediumvex, MIT. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file
Just SKILL.md in skills/09-ai-avatar of rediumvex/ai-video-generator-claude.
Open the folder on GitHubat commit ffdad7d
Seedance AI Avatar next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.
| Skill | Stars | Used in | Tokens | Auto-check | Licence | Repo updated |
|---|---|---|---|---|---|---|
| Seedance AI Avatar this skillrediumvex/ai-video-generator-claude | 409 | — | ~6.7k | Automated safety check: Pass | MIT | |
| AI Social Media ContentNeverSight/learn-skills.dev | 217 | 1 repos | ~1.8k | Automated safety check: Pass | None | |
| AI Marketing VideosNeverSight/learn-skills.dev | 217 | 3 repos | ~2.1k | Automated safety check: Pass | None | |
| Pexo Agentpexoai/pexo-skills | 804 | — | ~5k | Automated safety check: Pass | MIT-0 | |
| Seedance Social Hookbeshuaxian/higgsfield-seedance2-jineng | 952 | — | ~21k | Automated safety check: Pass | None | |
| Seedance Social Hookbeshuaxian/higgsfield-seedance2-jineng | 952 | — | ~20k | Automated safety check: Pass | None |
NeverSight/learn-skills.dev
Create AI-powered social media content for TikTok, Instagram, YouTube, Twitter/X.
NeverSight/learn-skills.dev
Create AI marketing videos for ads, promos, product launches, and brand content.
pexoai/pexo-skills
AI video generation skill with auto model selection across Seedance 2, Kling 3.0, HappyHorse, and 10+ models.
beshuaxian/higgsfield-seedance2-jineng
Generate viral social media hook video prompts for TikTok, Instagram Reels, and YouTube Shorts using Seedance 2.0 on Higgsfield.
beshuaxian/higgsfield-seedance2-jineng
使用 Seedance 2.0(Higgsfield)为 TikTok、Instagram Reels 和 YouTube Shorts 生成病毒式社交媒体钩子视频提示。在用户想要创建令人停下滚动的钩子、病毒式短视频、引人注目的开场白、TikTok 内容、Reels 内容、Shorts 内容或任何社交媒体优化视频时使用。在以下情况下触发:社交媒体视频、TikTok、Instagram…
aiskillstore/marketplace
AI video generation skill with auto model selection across Seedance 2, Kling 3.0, HappyHorse, and 10+ models.
rediumvex/ai-video-generator-claude
Generate before-and-after transformation video prompts for Seedance 2.0 on Higgsfield.
rediumvex/ai-video-generator-claude
Generate faceless content video prompts for Seedance 2.0 on Higgsfield.
rediumvex/ai-video-generator-claude
Generate personal brand and founder story video prompts for Seedance 2.0 on Higgsfield.
rediumvex/ai-video-generator-claude
Generate SaaS product launch and software demo video prompts for Seedance 2.0 on Higgsfield.
rediumvex/ai-video-generator-claude
Generate scroll-stopping viral hook video prompts for Seedance 2.0 on Higgsfield.
rediumvex/ai-video-generator-claude
Generate online course and coaching program promotional video prompts for Seedance 2.0 on Higgsfield.
Categories
Generate AI avatar and digital persona video prompts for Seedance 2.0 on Higgsfield. Seedance AI Avatar is an agent skill from rediumvex/ai-video-generator-claude.0 on Higgsfield.
Seedance AI Avatar fits situations like: virtual spokesperson content; digital twin videos; AI presenter clips; avatar-based marketing.
Run `npx skills add rediumvex/ai-video-generator-claude --skill seedance-ai-avatar -a claude-code`. Or copy the skill folder (skills/09-ai-avatar in rediumvex/ai-video-generator-claude) into .claude/skills/seedance-ai-avatar in your project. Claude Code loads it when a task matches its description.
Run `npx skills add rediumvex/ai-video-generator-claude --skill seedance-ai-avatar -a codex`. Or copy the skill folder (skills/09-ai-avatar in rediumvex/ai-video-generator-claude) into .agents/skills/seedance-ai-avatar in your project. Codex loads it when a task matches its description.
Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add rediumvex/ai-video-generator-claude --skill seedance-ai-avatar -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/seedance-ai-avatar, .gemini/skills/seedance-ai-avatar, .github/skills/seedance-ai-avatar and .opencode/skills/seedance-ai-avatar in your project.
SKILL.md names no scripts, command-line tools or credentials: Seedance AI Avatar is instructions for the agent only.
SKILL.md contains no URLs. Any network use would come from the scripts or tools the agent runs. This is read from the text; nothing was executed.
Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.
Seedance AI Avatar is published under the MIT licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.
About 6.7k tokens (SKILL.md is roughly 27k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.
Skills that share tags, products or a category with Seedance AI Avatar: AI Social Media Content (NeverSight/learn-skills.dev, 217 stars), AI Marketing Videos (NeverSight/learn-skills.dev, 217 stars), Pexo Agent (pexoai/pexo-skills, 804 stars) and Seedance Social Hook (beshuaxian/higgsfield-seedance2-jineng, 952 stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.
rediumvex (a GitHub user) maintains it in rediumvex/ai-video-generator-claude, which has 409 GitHub stars. The repository holds 10 skills in this directory. The repository was last updated on August 11, 2026.
Source: rediumvex/ai-video-generator-claude on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.