Agent Skills

social-graphics

Produce platform-correct social media graphics: pick the right dimensions and layout per platform, write the headline/body/CTA copy, build the generation prompt, and generate the image via the repo's FAL.ai pipeline. Use whenever the user wants a social post visual, Instagram post/story/carousel, LinkedIn graphic, Twitter/X image, TikTok thumbnail, Pinterest pin, launch announcement graphic, or 'something to post about X' — even if they never name a platform. Also use to adapt one design across

Install

npx skills add https://github.com/sanky369/vibe-building-skills --skill social-graphics
SKILL.md

Social Graphics

Turn a message and a platform into ready-to-post graphics. For each asset you produce a Social Graphic Spec — platform, exact dimensions, layout, overlay copy, and a complete generation prompt — then generate the image through the repo pipeline (docs/creative_cli.py, model fal-ai/nano-banana-pro). The governing principle: a social graphic is read for under two seconds — one message, one focal point, one CTA; if a viewer can't get the point from a thumbnail-sized glance, the design fails regardless of how good it looks full-size.

When to use / when not to

  • Use for platform-bound static visuals: feed posts, stories, pins, link-preview images, announcement graphics, quote cards, carousels.
  • Logos, icons, patterns, brand marks → skills/creative/brand-asset.
  • Standalone product hero shots (no platform framing, no overlay copy) → skills/creative/product-photography.
  • Motion content → skills/creative/product-video (product-led) or skills/creative/talking-head (person-led).
  • No brand style yet and the user wants a consistent look → run skills/creative/creative-strategist first; its style guide feeds this skill.
  • If the visual language itself feels generic and needs an original identity → skills/creative/original-design first.

Intake

Ask in one tight batch, only what's missing:

  1. Message — what is this post saying, in one sentence? What should the viewer do (CTA)?
  2. Platform(s) and format — Instagram feed/story/carousel, LinkedIn, Twitter/X, TikTok, Pinterest? One platform or a multi-platform set?
  3. Brand constraints — style guide from skills/creative/creative-strategist, brand colors/fonts, product imagery to feature?
  4. Tone — professional, trendy, minimalist, playful?

Infer instead of asking: tone from the platform (LinkedIn → professional; TikTok → energetic), colors from any brand materials in the project. Don't stall: with a clear message and platform, state assumptions for the rest and proceed.

Platform specifications (source of truth for dimensions)

Platform Format Pixels CLI aspect-ratio Design posture
Instagram Feed post 1080×1080 1:1 vibrant, high contrast, trending aesthetic
Instagram Story / Reel cover 1080×1920 9:16 full-bleed vertical, text in safe zone
Instagram Carousel 1080×1350 4:5 one idea per card, visual continuity across cards
LinkedIn Feed post 1200×627 16:9* professional, credible, value-led copy
LinkedIn Document page 1200×1500 4:5 clean, readable, business imagery
Twitter/X Post image 1200×675 16:9 bold, simple, instant visual impact
TikTok Thumbnail/graphic 1080×1920 9:16 bold text overlays, energetic, authentic feel
Pinterest Standard pin 1000×1500 2:3 vertical, inspiring, high-quality imagery

*Closest supported ratio; the CLI supports 21:9, 16:9, 3:2, 4:3, 5:4, 1:1, 4:5, 3:4, 2:3, 9:16 — always pick the nearest and note any crop.

Workflow

1. Resolve platform and format

If the user named a platform, use the table above. If not, decide: B2B/professional audience → LinkedIn; consumer/lifestyle → Instagram; discovery/how-to/aspirational → Pinterest; announcement to an existing following → the platform where they have one (ask only if truly unknown). Multi-platform request → design the master for the most constrained format first (usually 1:1 or 9:16), then adapt per platform; never stretch one image across all.

2. Write the copy before the visual

  • Headline: ≤10 words, the message itself, not a teaser.
  • Body (optional): ≤2 short lines supporting the headline.
  • CTA: 2–4 words, action verb ("Shop now", "Get the guide", "Link in bio").

If the copy can't fit these limits, the post is carrying more than one message — split it into a carousel or multiple posts and tell the user why.

3. Design the layout

Pick one focal point (product shot, illustration, bold typography) and place copy against the quietest area. Decision rules: photo-led → headline top or bottom third over a darkened/blurred band; typography-led → headline IS the focal point, minimal imagery; carousel → identical layout grid on every card, only content changes. Rule of thirds, generous negative space, max 2–3 font roles (headline / body / CTA).

4. Build the generation prompt

Formula — every prompt contains all six parts, in order:

[Platform + format] graphic, [subject/focal point], [color palette from brand or chosen],
[composition: focal point + where copy space is reserved], [tone/mood words],
[dimensions], professional design, high contrast, eye-catching

Always reserve explicit "clean negative space at [position] for text overlay" in the prompt — nano-banana-pro renders text imperfectly, so plan to overlay final copy in a design tool rather than trusting generated text for anything longer than 2–3 words.

5. Generate

If FAL_API_KEY (or FAL_KEY) is set, run the repo CLI (docs/fal_api.py under the hood, model fal-ai/nano-banana-pro):

python docs/creative_cli.py social \
  --platform instagram --topic "Product Launch" \
  --prompt "<full prompt>" --aspect-ratio 1:1 --resolution 2K

--platform accepts instagram|linkedin|twitter|tiktok|pinterest. Generate --num-images 2 when the user is choosing between directions. If no key is set, deliver the spec with ready-to-run commands instead.

6. Self-check against the quality bar, deliver the spec + images.

Required output format

One block per asset; a multi-platform set is the master followed by adaptation blocks:

# Social Graphic Spec — [Campaign/Message]

## Asset 1 — [Platform · Format]
- **Dimensions:** 1080×1080 (aspect `1:1`) · **Role:** master | adaptation
- **Message:** [one sentence]

### Copy
- Headline: "..." (≤10 words)
- Body: "..." (≤2 lines, or —)
- CTA: "..."

### Layout
- Focal point: ... · Copy placement: ... · Palette: [colors + source: brand guide / chosen]

### Generation prompt
> [full prompt, all six parts]

`python docs/creative_cli.py social --platform [x] --topic "[x]" --prompt "..." --aspect-ratio [x] --resolution 2K`

### Result
- [generated file path(s), or "pending — FAL_API_KEY not set"]
- Post-gen step: overlay final copy in design tool: [yes/no + what]

## Asset 2 — [Platform · Format]
(same structure; for adaptations, note only what changes from the master)

Quality bar (check before delivering)

  • Dimensions match the platform table exactly; aspect ratio is a supported CLI value.
  • Headline ≤10 words; total text ≤3 elements (headline, body, CTA); one message per asset.
  • Copy placement sits over reserved negative space, not over the focal point.
  • Text-background contrast is explicitly handled (band, shadow, or blur called out in layout).
  • Palette and tone match the brand style guide when one exists; deviations flagged to the user.
  • Prompt contains all six formula parts, including reserved copy space.
  • Multi-platform sets: adapted per format, never one image stretched; carousel cards share one grid.
  • No invented engagement statistics; "best practice" guidance here is heuristic — label it if asked.

Integration

  • skills/creative/creative-strategist → feeds in: the style guide (palette, typography, mood) every spec must obey.
  • skills/creative/original-design → feeds in: a design-language brief when the user wants an original visual world; every graphic becomes an artifact of it.
  • skills/creative/product-photography → feeds in: product shots reused as focal-point imagery.
  • skills/creative/product-video → feeds in: hero keyframes reused as post visuals; consumes: nothing.
  • skills/creative/brand-asset → feeds in: logos/marks to place; consumes: nothing.
  • skills/creative/orchestrator → routes multi-asset campaigns through this skill per platform.

Related skills

ai-image-generationgenmedia-labs713KGenerate and edit images on RunComfy via the `runcomfy` CLI — a smart router across the full image-model catalog: FLUX 2 (Klein 9B/4B, Pro, Dev, Flash, Turbo, Max), Google Nano Banana 2 / Pro, OpenAI GPT Image 2, ByteDance Seedream 5 / 4-5 / 4-0 and Dreamina 4-0, Alibaba Qwen Image and Z-Image Turbo, Wan 2-7. Covers both text-to-image (t2i) and image-to-image / edit (i2i) endpoints — the skill picks the right model for the user's actual intent (typography precision, photoreal portraits, sub-secoai-image-generation101-skills547KGenerate AI images with GPT-Image-2, FLUX, Gemini, Grok, Seedream, Reve and 50+ models via inference.sh CLI. Models: GPT-Image-2, FLUX Dev LoRA, FLUX.2 Klein LoRA, Gemini 3 Pro Image, Grok Imagine, Seedream 4.5, Reve, ImagineArt. Capabilities: text-to-image, image-to-image, inpainting, LoRA, image editing, upscaling, text rendering. Use for: AI art, product mockups, concept art, social media graphics, marketing visuals, illustrations. Triggers: flux, image generation, ai image, text to image, stnano-banana-2prime-skills424KGenerate images with Google Nano Banana 2 (Gemini-family flash-tier text-to-image) on RunComfy — bundled with the model's documented prompting patterns so the skill gets sharper output than naive prompting against the same model. Documents Nano Banana 2's strengths (rapid iteration, in-image typography rendering, predictable framing, optional web-grounded context), the resolution-tier pricing, the safety-tolerance dial, and when to route to Nano Banana Pro / GPT Image 2 / Flux 2 / Seedream insteimage-editprime-skills424KEdit images on RunComfy — this skill is a smart router that matches the user's intent to the right edit model in the RunComfy catalog. Picks Nano Banana Edit (batch up to 20, identity-preserving default), OpenAI GPT Image 2 Edit (multilingual in-image text rewrite, multi-ref composition, layout precision), Flux Kontext Pro (single-ref high-fidelity local edit), or Z-Image Turbo Inpaint (mask-driven precise region edit). Bundles each model's documented prompting patterns so the skill gets sharper

Search skills and MCP servers

Fuzzy search across 23,137 skills and servers