Agentic Hub

Viral Creative Rewrite Skill with byteplus's seedance 2.0 model helps advertisers turn a reference ad into a new product ad video instead of starting from scratch. The reference ad provides a proven creative structure, including the opening hook, pacing, shot order, product moments, satisfaction points, and ending rhythm, while the advertiser’s product image provides the source of truth for the new product. After the user invokes the skill, it asks them to choose a flow, select default or custom media, and provide product identity, must-keep selling points, target audience, and campaign goal. The skill supports a no-cost rehearsal with built-in example assets, so advertisers can review the guided flow, rewrite brief, detailed analysis, and example result before using real generation. For real use, the skill first prepares an analysis preview. It explains what can be borrowed from the reference ad, what must not be inherited, how the product image anchors the new video, output settings, and risk controls before any video generation step. Final video generation only happens after the user reviews and confirms the prepared brief. This helps ecommerce and performance advertisers adapt a proven ad structure to their own product while avoiding inherited brands, unsupported claims, captions, UI elements, or unrelated template products. This skill was created through a conversational Claude Code skill workflow and defaults to Claude 4.7 as the host agent. The ZIP includes built-in rehearsal and example video assets; if a standard upload flow enforces a 30 MB file-size limit, the same skill can be registered through the Claude skill path or submitted as a size-adjusted package with large example assets externalized.

Tldr: Read-only TikTok creative engine: rank winners, verify claims against rejection history, or build a cold-start packet. tt4b-creative-engine is a read-only creative decision engine for TikTok advertisers, built on the tt-ads-mcp-flat server. It turns "what should we run next" from guesswork into an evidence chain, covering both ends of the official Advertiser-Maturity axis. Mode A (accounts with delivery history — Experienced & Scaling) pulls real performance reports to rank winning ads by hook-rate, CTR, and conversion_rate; infers the winning ad's structure from retention curves and ad copy with confidence-tagged hypotheses; cross-checks every product claim against the advertiser's own catalog and ad-rejection history (a Do-Not-Say list no generation tool can produce, because it requires the account's actual rejection records); and delivers a claim-safe replication brief plus an A/B test plan with MDE-derived thresholds and arm prioritization. It replicates structure and strategy — never clones or generates video. Mode B (brand-new accounts — New to TikTok Ads / Emerging Market) builds a cold-start brief from live market data: keyword search volume, CPC ranges, audience-size estimates, and interest insights. It closes with a schema-validated, dry-run campaign starter packet carrying a [DRY_RUN_PREVIEW] marker, field-by-field source tagging, and a hand-off line to the official tt4b-launch-* skills. This addresses three concrete advertiser pains: not knowing what to shoot next, blank-page paralysis on new accounts, and creative claims that get ads rejected. The judged path is fully read-only: zero write or spend calls, enforced as a one-strike eval invariant.

The Ad Group Optimizer is an intelligent performance optimization agent that analyzes ad group performance and recommends or implements bid, budget, and targeting optimizations based on data-driven thresholds. This skill automates the complex process of identifying top performers, scaling opportunities, and underperforming ad groups while providing actionable recommendations backed by TikTok's diagnostic insights. It segments campaigns into performance categories, calculates optimization scores, and can automatically implement budget adjustments, bid modifications, and targeting refinements—ensuring your campaigns stay competitive while maximizing ROAS and minimizing CPA. Key Capabilities: Multi-dimensional performance analysis across CPA, ROAS, CTR, and conversion metrics Automated segmentation into Top Performers, Scaling Opportunities, Underperformers, and Testing Phase Integration with TikTok's native diagnostic tools for platform-specific recommendations Safe auto-implementation with built-in guardrails (2x spend caps, minimum data thresholds) Before/after impact forecasting with monitoring period recommendations

Evaluates email campaigns for alignment with industry regulations, deliverability risks, accessibility guidelines, and best practices.

This Skill helps TikTok performance / Brand Ads advertisers generate launch-ready Full-Funnel ad recipes in one click — based on brand/performance goals, budget, market, and existing assets — covering all four funnel layers: Awareness → Consideration → Action → Retention. It includes: audience diagnosis → recommended product combinations → budget & frequency allocation → creative strategy → showcase strategy recommendations. Key differentiator: rather than advising on a single product, it orchestrates cross-product combinations; it optimizes for funnel incrementality and ROAS / BUR / CPA improvement rather than a single KPI.

This skill helps diagnose whether a C&S Newbie advertiser or campaign follows key best practices for creative quantity, budget, and bid setup.

Sync Shoplazza catalog to TikTok Ads, create smart+ A/B test and analyze the results.

TikTok pre-launch audience analysis playbook. Takes a landing page URL as input and delivers a complete pre-advertising strategy: product positioning analysis (6 dimensions), target user persona with demographics, competitor landscape with budget recommendations, target region selection, and 5 deployable audience profiles with TikTok-validated interest keywords, gender ratios, and age ranges. Covers TikTok Campaign Objective two-layer mapping, sensitive industry policy detection, and a dual-layer interest keyword mining pipeline with 4-perspective seed word construction.

Account-Specific Diagnosis Copilot is a context-aware AI assistant purpose-built for advertising account troubleshooting. Rather than offering generic optimization tips, it delivers prioritized, evidence-backed diagnosis by: Detecting what changed — It identifies recent shifts in account or campaign status, surfacing the specific variables that moved. Explaining why performance dropped — It performs root-cause analysis using historical performance baselines and recent anomaly signals to pinpoint the drivers behind under-delivery, cost spikes, or revenue dips. Recommending what to do right now — It produces actionable next steps ranked by impact, so operators can act immediately rather than sift through dashboards. It pulls from account-level context — including campaign configuration, historical trends, and real-time anomalies — to build a complete evidence chain from symptom → root cause → fix. This makes it ideal for advertiser-facing or ops teams that need fast, precise answers instead of broad playbook advice.