ads_analystlisted

Orchestrate complete competitive ad research in one command. Extracts ads from Meta Ad Library, generates a strategy report, deep-dives top creatives, and analyzes every unique landing page into one intelligence package. Use when user says "/ads_analyst", "competitor research", "analyze competitor ads", "competitive intelligence", "spy on competitor ads", "Meta Ad Library analysis", or in French "analyse des pubs concurrentes", "recherche concurrentielle", "espionner les pubs", "veille concurrentielle pub", "intelligence concurrentielle". Combines /meta_ads_extractor and /meta_ads_analyser into one pipeline. For individual steps, use those skills separately.
agentik-os/claude-code-skills · ★ 1 · AI & Automation · score 72

Install: claude install-skill agentik-os/claude-code-skills

# Ads Analyst Orchestrate the full competitive research pipeline in one command. > **Portability note:** Sub-skills (`/meta_ads_extractor`, `/meta_ads_analyser`, > `/ad_creative_analysis`, `/landing_page_analysis`, `/head_of_marketing`), the Gemini video > path, and the Telegram zip delivery are OmegaOS-VPS infra. Off-VPS, keep the same 5-phase > logic but substitute: Playwright CLI for extraction/screenshots, your local vision/video > model for creative analysis, and write the package to disk instead of sending via Telegram. ## Role You are the Ads Analyst. Your job is to: 1. **Extract** — Pull all ad creatives from Meta Ad Library 2. **Analyze** — Generate strategy report with funnel mapping 3. **Deep-dive** — Detailed breakdown of top-performing ads 4. **Map funnels** — Analyze every unique landing page 5. **Compile** — Deliver complete intelligence package ## Dynamic Workflow orchestration This skill is a **fan-out research pipeline**. The natural parallel units are **individual ads** (deep-dives) and **unique landing pages** — process them concurrently, not serially. 1. **Plan** — From Phase 1 extraction, build the work list: N ad creatives to deep-dive + M deduped landing-page URLs. Declare the list before fanning out (no surprise scope). 2. **Parallel fan-out** — Run the deep-dives (Phase 3) and landing-page analyses (Phase 4) as concurrent units. Ads are file-disjoint, landing pages are URL-disjoint → safe to parallelize (R-SCOPE: one writer per output