# Fashion Campaign Production > Provider-independent production workflow for AI-assisted fashion campaigns, lookbooks, editorial fashion films, ecommerce/editorial hybrids, social cutdowns, virtual try-on assets, generated model/lifestyle ads, and retouched fashion visuals; use when an agent must plan, prompt, produce, review, localize, or deliver fashion imagery or video while preserving garment truthfulness, model consent, fi… - **Type:** Skill - **Install:** `agentstack add skill-calesthio-generative-media-skills-fashion-campaign-production` - **Verified:** Yes — security-reviewed for prompt injection and unsafe behavior - **Seller:** [calesthio](https://agentstack.voostack.com/s/calesthio) - **Installs:** 0 - **Category:** [Developer Tools](https://agentstack.voostack.com/c/developer-tools) - **Latest version:** 0.1.0 - **License:** MIT - **Upstream author:** [calesthio](https://github.com/calesthio) - **Source:** https://github.com/calesthio/generative-media-skills/tree/main/skills/production/content-formats/fashion-campaign-production - **Website:** https://github.com/calesthio/OpenMontage ## Install ```sh agentstack add skill-calesthio-generative-media-skills-fashion-campaign-production ``` Requires the [AgentStack CLI](https://agentstack.voostack.com/docs/cli). Works with Claude Code, Cursor, and any MCP-compatible agent. ## About # Fashion Campaign Production Use this skill to make fashion content that is beautiful, commercially useful, and honest about the garments, people, claims, and media provenance. Treat fashion production as a truth-preserving workflow: the campaign can stylize mood, setting, casting, edit, and story, but it must not invent product attributes, misrepresent fit, fake substantiation, or use a person's likeness without permission. This is not legal advice. When a claim, disclosure, release, marketplace decision, or jurisdictional rule affects real publication, ask the client, counsel, marketplace lead, or platform specialist to approve before launch. ## Production principle The garment is the source of truth. Everything else exists to help a customer, editor, buyer, fan, or viewer understand why that garment matters. For every asset, distinguish: - Documented facts: SKU, silhouette, fiber content, colorway, finish, construction, care, country of origin, size range, price, availability, tested performance, sustainability claims, endorsements, model releases, platform rules, and delivery specs. - Creative choices: location, lighting, mood, casting archetype, pose, camera language, pacing, typography, and music. - Production heuristics: fit representation, retouching limits, color checks, garment-detail closeups, prompt repair tactics, and QA routines. Do not let creative choices override documented facts. ## Intake: ask for the campaign truth packet Before generating or editing fashion visuals, collect or infer enough of this packet to proceed safely. If information is missing, make a visible assumption for low-risk creative choices and pause for high-risk facts. ### Brand and campaign brief - Brand, collection, season, drop date, markets, target platforms, paid/organic use, and campaign objective. - Brand codes: luxury, streetwear, minimal, craft, performance, archival, inclusive, avant-garde, sustainable, sensual, athletic, youth, mature, etc. - Audience and context: editorial awareness, conversion, wholesale buyer deck, product detail page, social launch, email, outdoor, in-store display, press kit, or runway recap. - Competitive references and forbidden references. Avoid copying another brand's campaign structure, photographer signature, runway look, model styling, typography, or distinctive trade dress. ### Garment truth sheet For each hero item or SKU, capture: - Product name/SKU, category, colorway, size range, availability status, and markets. - Silhouette, fit, length, rise, sleeve shape, neckline, closure, pocketing, lining, hardware, seam/stitch details, trims, logo placement, print placement, wash, finish, transparency, stretch, drape, and intended layering. - Fabric/fiber/material claims, care instructions, country-of-origin claims, and any regulated material claims such as wool, fur, leather, recycled fibers, organic fibers, or performance finishes. - Must-show details and must-not-show angles. - Verified reference images: flat lay, on-model, back view, macro details, color swatch, label, hangtag, CAD/tech pack, or approved packshot. If the sheet is incomplete, avoid exact claims and avoid close visual details that the system may hallucinate. For ecommerce or marketplace use, do not create a "product photo" from imagination unless the client explicitly approves the asset as illustrative or concept-only. ### People, likeness, and releases Capture: - Whether the model is a real person, licensed stock person, influencer, employee, synthetic person, public figure, or composite. - Release scope: brand, product line, media channels, territories, duration, paid ads, edits/retouching, AI training or generation, digital replica, voice, body scan, face replacement, and derivative cutdowns. - Whether generated or edited content resembles an identifiable person. - Restrictions: no face, no body morphing, no tattoos, no visible age cues, no implied ethnicity, no pregnancy implication, no religious garments, no sensitive attributes, no minors, no sexualized styling, etc. Escalate when releases are missing, scope is ambiguous, or the prompt asks to resemble a celebrity, influencer, real model, customer, employee, private person, or a recognizable campaign image. ## Risk triage before production Pause for client/counsel/marketplace approval if the work includes any of the following: - Sustainability, environmental, ethical, vegan, cruelty-free, recycled, organic, carbon, water-saving, fair labor, local-made, "Made in USA," antimicrobial, UV-protection, slimming, shapewear performance, posture, health, or safety claims. - Textile, wool, fur, care-label, country-of-origin, or fiber-content statements that will appear in commerce. - Influencer, testimonial, UGC, affiliate, gifted product, employee endorsement, or "real customer" claims. - Real-person likeness, digital replica, synthetic performer, face swap, voice clone, public figure, minors, or sensitive personal attributes. - Marketplace or paid ad use on platforms with active review policies. - International launch, regulated categories, culturally specific garments, religious dress, uniforms, protected designs, or brand collaborations. - Body-image-sensitive concepts: before/after, "fix your body," weight loss, anti-aging, skin-lightening, "perfect body," or body-shaming copy/visuals. If risk is high and approval is missing, produce a safe concept deck, storyboard, prompt plan, or QA checklist instead of final publishable assets. ## Build the fashion concept Develop the concept from the collection outward: 1. Identify the collection tension: restraint vs movement, tailoring vs softness, archive vs future, utility vs ornament, resort ease vs city armor, atelier craft vs everyday wear. 2. Choose the campaign mode: - Lookbook: systematic garment coverage with consistent styling, poses, and lighting. - Editorial: story, attitude, and art direction can dominate, but product truth remains visible. - Fashion film: movement, texture, pacing, gesture, and music carry the collection world. - Ecommerce/editorial hybrid: product information and conversion clarity coexist with richer campaign atmosphere. - Collection launch/social: high-recognition visual hook, fast first read, modular cutdowns. - Virtual try-on or fit assets: controlled, measurement-aware, neutral enough to preserve fit expectations. 3. Define the viewer promise: "see the drape," "feel the tailoring," "understand the drop," "want the lifestyle," "trust the fit," or "remember the collection world." 4. Map hero products to shots. Do not give every item the same status; decide which garments carry story, proof, texture, movement, and conversion. ### Fashion shot grammar Use the shot list to balance desire and information: - Establishing world: location, season, attitude, palette. - Full look: head-to-toe silhouette and proportions. - Fit proof: front, side, back, seated, walking, reaching, or layered movement. - Detail macro: stitch, hardware, closure, texture, label, hem, wash, print alignment. - Fabric behavior: drape, stretch, sheerness, reflection, pile, compression, crinkle, weight. - Styling beat: shoes, bag, jewelry, hair/makeup, but avoid overwhelming hero garments. - Social hook: one frame that reads in one second without distorting the garment. - Product-safe frame: clean crop usable for PDP, paid social, email, or thumbnail. For video, write beats rather than generic mood: - 0-2s: instant collection signal and hero silhouette. - 2-5s: motion or gesture proves fabric/fit. - 5-8s: detail or styling beat. - 8-12s: second look or contrast. - 12-15s: logo/drop/date/CTA, if approved. ## Garment truthfulness rules Do: - Anchor image/video prompts to verified reference assets, SKU facts, and a "do not alter" list. - Preserve the garment's silhouette, neckline, sleeve, hem, closure, pocket, logo, print, pattern scale, color family, texture, and material behavior. - Include neutral fit documentation when the asset is for ecommerce, sizing, VTO, marketplace listings, or customer expectation. - Keep exact product claims out of generated backgrounds, captions, and overlays unless substantiated. - Label concept renders, inspirational frames, or synthetic model images internally so downstream users do not mistake them for documentary product photos. Do not: - Invent garment construction, fabric composition, certifications, origin, care instructions, performance properties, availability, price, collaborator, or sustainability benefits. - Use a generated garment as a final product image when shoppers could believe it is the actual item. - Hide fit issues with poses, clips, clamps, impossible drape, extreme retouching, or fabric physics that the real garment cannot achieve. - Change body proportions to make the item "fit better." - Generate fake tags, labels, certification seals, QR codes, care icons, designer signatures, or marketplace badges. - Show unavailable colorways, sizes, bundles, discounts, or variants as purchasable. ## Styling and art direction Style to clarify the collection, not to rescue a weak prompt. - Keep silhouette hierarchy clear. If the coat is the hero, do not bury it under scarves, bags, props, or cropped framing. - Choose underlayers that explain transparency, proportion, and season. - Use accessories to support the brand world, not to imply partnerships or luxury affiliations that do not exist. - Avoid logo confusion: do not place a generated garment next to recognizable third-party marks unless licensed and intentional. - Watch cultural garments, religious dress, uniforms, indigenous patterns, and heritage motifs. If they are central to the brief, request cultural review and sourcing context. - Make makeup, hair, skin finish, and pose compatible with the product category: tailoring needs structure; knitwear needs touch and softness; performance wear needs movement proof; eveningwear needs light behavior; denim needs wash and fit details. ## Model, fit, body, and retouching ethics Fashion visuals shape body expectations. Treat model representation as part of product truth. - Specify model diversity intentionally: body size, age range, skin tone, mobility, gender presentation, and styling context when relevant to the brand and audience. - Do not imply the viewer has a personal attribute, defect, medical condition, body problem, or identity trait. Phrase copy around garment benefits and styling, not "fixing" the audience. - Do not body-morph models, slim waists, lengthen legs, alter age, lighten/darken skin, erase disability aids, or change protected traits unless the person and client have explicitly approved a specific edit for a legitimate production reason. - Retouch skin and fabric differently: remove transient artifacts such as lint, sensor dust, or stray threads only when they are not product features; preserve skin texture and garment texture. - For fit-relevant imagery, keep natural folds, tension points, hems, and movement. Excessive smoothing can mislead shoppers. - For plus-size, adaptive, maternity, petite, tall, or gender-inclusive lines, include fit evidence rather than only mood crops. - For virtual try-on, avoid beauty retouching beyond normal display correction. The asset should support fit interpretation, not fantasy transformation. ## Claims, endorsements, and substantiation Documented fact: The FTC's advertising substantiation policy expects advertisers to have a reasonable basis for objective claims before an ad is disseminated; the FTC also treats both express and implied claims as relevant. Verified 2026-07-11: https://www.ftc.gov/legal-library/browse/ftc-policy-statement-regarding-advertising-substantiation Apply this production rule: - Turn every visible or implied claim into a claim ledger entry. - Ask what a reasonable viewer could infer from the image, copy, voiceover, overlay, landing page, product title, and influencer context together. - Require substantiation before using objective claims such as "waterproof," "all-day compression," "made from recycled ocean plastic," "organic cotton," "carbon neutral," "ethically made," "fits every body," "Made in USA," "wrinkle-free," or "UV protection." - If substantiation is missing, rewrite to subjective or non-claim language: "crisp handfeel," "structured silhouette," "designed for layering," "cool-weather styling," "recycled-content story pending certification review." Documented fact: FTC Green Guides are intended to help marketers avoid misleading environmental claims. Verified 2026-07-11: https://www.ftc.gov/news-events/topics/truth-advertising/green-guides For sustainability and ethics claims: - Avoid broad unqualified terms like "sustainable," "eco-friendly," "green," "ethical," or "conscious" unless counsel/client approves evidence and qualification. - Prefer specific, scoped, evidenced claims: material percentage, certification, supply-chain boundary, product part, year, and method. - Do not imply the entire garment or brand is sustainable because one trim, packaging element, or capsule uses a preferred material. - Keep claim evidence with the asset package: certificates, supplier declarations, test reports, LCA summaries, or approved legal copy. Documented fact: FTC guidance for textile/wool labeling says most textile and wool products need labels with fiber content, country of origin, and responsible company identity; care labels for clothing are handled by a separate FTC rule. Verified 2026-07-11: https://www.ftc.gov/business-guidance/resources/threading-your-way-through-labeling-requirements-under-textile-wool-acts Documented fact: FTC care-label guidance says the Care Labeling Rule requires manufacturers and importers to attach care instructions to clothing and some piece goods. Verified 2026-07-11: https://www.ftc.gov/business-guidance/resources/clothes-captioning-complying-care-labeling-rule Documented fact: FTC Made in USA guidance says the FTC does not pre-approve Made in USA claims; marketers must ensure claims are truthful and substantiated. Verified 2026-07-11: https://www.ftc.gov/business-guidance/resources/complying-made-usa-standard Documented fact: FTC endorsement guidance says endorsements must be honest, not misleading, and material connections that consumers would not expect should be clearly and conspicuously disclosed. Verified 2026-07-11: https://www.ftc.gov/business-guidance/resources/ftcs-endorsement-guides-what-people-are-asking For influencers, UGC, stylists, employees, and "real customer" creative: - Include disclosure copy in the deliverable plan, not as an afterthought. - Do not script an influencer to make a claim the brand cannot make. - Do not fabricate UGC, comments, reviews, try-on reactions, or "street style" documentation. - If using AI to create simulated UGC, label it internally and seek platform/legal approval before any public use. ## Synthetic models, digital replicas, and provenance Use synthetic people only when the campaign can be honest about them and when the brand has approved that representation. - Do not prompt for a real model, celebrity, influencer, public figure, customer, employee, or private person without explicit rights. - If a synthetic person is intended to look human but not like an identifiable person, document it as a synthetic performer/model and check disclosure requirements for each launch market. - If a real model's face/body/voice is scanned, cloned, face-swapped, or extended into new performances, require release scope for digital replica use and generated derivatives. - Preserve provenance metadata when possible. Maintain a creation log with tool, model/version, prompt, references, seed if available, edit chain, retouch notes, approvals, and disclosure status. Documented fact: C2PA describes Content Credentials as a content provenance/authenticity standard that can show information about media history. Verified 2026-07-11: https://c2pa.org/ Documented fact: … ## Source & license This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code. - **Author:** [calesthio](https://github.com/calesthio) - **Source:** [calesthio/generative-media-skills](https://github.com/calesthio/generative-media-skills) - **License:** MIT - **Homepage:** https://github.com/calesthio/OpenMontage Install and usage instructions live in the source repository linked above. ## Pricing - **Free** — Free ## Security capabilities Automated source analysis of v0.1.0 — what this tool can access: - **Network access:** no - **Filesystem access:** no - **Shell / process execution:** no - **Environment & secrets:** no - **Dynamic code execution:** no *"Yes" means the capability is present in the source — more access means more to trust, not that it is unsafe.* ## Versions - **0.1.0** — security scan: passed — Imported from the upstream source. ## Links - Listing page: https://agentstack.voostack.com/l/skill-calesthio-generative-media-skills-fashion-campaign-production - Seller: https://agentstack.voostack.com/s/calesthio - Browse the marketplace: https://agentstack.voostack.com/browse --- Listed on AgentStack — the marketplace for AI agent skills and MCP servers. Every listing is security-reviewed. Creators keep 70%.