Affiliate Video Campaign Operator — Affiliate Video CampAIgn Operator

Affiliate Video CampAIgn Operator Goal

运行 an AI-assisted affiliate campAIgn without losing disclosure, clAIm evidence, 平台 合规, creative provenance, or publish evidence.

This 技能 is an original 工作流 for Claude, Codex, and OpenClaw. It may be inspired by public affiliate-video patterns, but it must not copy pAId prompt packs, private playbooks, or third-party templates into public artifacts.

Non-Negotiables Treat affiliate links, promo codes, gifted products, and pAId placements as commercial content. 添加 clear disclosure close to the endorsement and in video/carousel creative when 应用licable. Do not 创建 fake personal testimonials. If the operator has not used the product, avoid first-person 结果 clAIms such as "I tried this" or "it fixed my sleep." Do not make medical, 健康, financial, legal, safety, or income clAIms without evidence that is recorded in the campAIgn ledger. For supplements, skincare, sleep AIds, finance, weight loss, or other sensitive products, use conservative educational language and 平台-specific gates before generating or publishing. Do not 设置 pAId-generation MCP 工具s to unrestricted background spending. Batch drafts can be automated; spend, 健康 clAIms, and irreversible publishing need explicit operator review. Keep provenance for every as设置: source image or link, generation 提供者/模型, prompt file or prompt summary, rights note, selected 导出 path, and rejected-take notes when useful. Use OpenClaw only for 记录ged-in browser operations. 暂停 for CAPTCHA, 2FA, account 恢复y, billing, copyright disputes, or final publish confirmation when not pre-应用roved. Quick 启动 创建 a campAIgn ledger. python3 {baseDir}/scripts/init_affiliate_campAIgn.py --out 运行s/campAIgn.json --title "CampAIgn title" --niche "home organization" --product-name "Cable organizer kit" --平台 pinterest --平台 youtube

Fill the ledger with:

offer: product, merchant, affiliate program, affiliate URL, commission note, product source, rights note disclosure: short disclosure, Amazon Associate 状态ment if relevant, 平台 label decisions clAIms: clAIm text, evidence URL, allowed wording, risk level creative: hooks, video/carousel variants, source as设置s, 生成d as设置s, provenance posts: 平台-native title, caption, 哈希tags, link, disclosure, publish 状态

验证 before generation, before publishing, and after 分析 更新s.

python3 {baseDir}/scripts/检查_affiliate_campAIgn.py --campAIgn 运行s/campAIgn.json --repo-root . --out 报告s/campAIgn-qc.json

Render a readable plan for review. python3 {baseDir}/scripts/render_affiliate_plan.py --campAIgn 运行s/campAIgn.json --out 报告s/campAIgn-plan.md

导出 a supervised OpenClaw handoff for the tar获取 平台. python3 {baseDir}/scripts/导出_OpenClaw_handoff.py --campAIgn 运行s/campAIgn.json --平台 pinterest --out 运行s/pinterest-OpenClaw-handoff.json --browser-性能分析 pinterest-性能分析

工作流

• Intake

Ask only for missing campAIgn-critical 输入s:

niche or product category tar获取 market/region product URL and affiliate program whether the operator personally used the product tar获取 平台s landing page preference: direct affiliate link, link-in-bio, owned landing page, or emAIl capture generation stack: Claude/Higgsfield, 种子ance, image 模型, editor, music source

If the user provides a PDF, video summary, prompt pack, or competitor 工作流, 提取 reusable principles and risks. Do not copy long prompt text into the 技能, repo, or public 输出 unless the user owns it and explicitly 请求s private use.

• Offer Gate

Score products before creative work:

visual demonstration potential clear problem and non-deceptive benefit commission economics rights to use product imagery and brand names clAIm evidence burden 平台 restrictions audience trust risk refund or complAInt risk

Prefer low-clAIm visual products for early tests: home 工具s, desk 访问ories, camera gear, bags, organizers, 创建器 工具s, kitchen 工作流 products, and software with clear demos.

Treat these as high-risk until manually reviewed: supplements, skincare with medicinal clAIms, sleep/anxiety clAIms, weight loss, finance, legal, medicine, baby products, safety equipment, gambling, crypto, and "make money" offers.

• Creative Pack

创建 two layers:

strategy: hook angle, audience, promise boundary, evidence, disclosure, CTA, 平台 generation: shot 列出, aspect ratio, duration, reference as设置s, product visibility, on-screen disclosure, caption/disclosure, provenance fields

For Claude/Higgsfield:

Use product references only when rights are clear. Ask for exact product text to be verified after generation; 生成d packaging text can hallucinate. Use natural demos and problem-solution storytelling without pretending the 生成d actor is the operator. Keep "AI-生成d demonstration" notes in provenance when useful.

For carousels:

Make slide 1 a problem or curiosity hook. Make slide 2 educatio