Audit Paid Ads Across All Platforms

Runs a comprehensive audit of paid ad accounts across all major platforms in 10-15 minutes, scoring account health and generating prioritized fixes. A manual audit of a single account takes 4-6 hours of senior PPC time and misses platform-specific blind spots; this skill deterministically covers 250+ checks across 8 platforms with local, verifiable rigor.

Once installed, you provide account context (industry, budget, active platforms) and account data (exports, screenshots, or metrics). The skill either dispatches 6 parallel audit agents (for full /ads audit) or runs a platform-specific deep dive (/ads google, /ads meta, etc.). Each audit generates a 0-100 health score with weighted per-platform breakdown, critical findings ranked by severity, and quick wins fixable in under 15 minutes. Output: structured JSON report plus optional PDF for client delivery.

Every paid ad account gets audited the same way with the same rigor every week, catching platform-specific issues (Andromeda creative suppression on Meta, AI Max negative keyword gaps on Google, ACOS/TACOS misalignment on Amazon) that human reviewers habitually miss because platforms change faster than audit checklists.

Requires Python 3.10+ because the eval harness uses pytest with modern type hints and the install matrix validates cross-host targets (Claude Code, Cursor, Windsurf, Codex, Gemini CLI, Goose) against a whitelist to prevent shell injection. PDF generation via reportlab is optional for /ads report. Landing page analysis via Playwright is optional for /ads landing. Image generation requires a provider key (Gemini, OpenAI, Stability, or Replicate) in .env; if missing, /ads generate and /ads photoshoot exit gracefully with setup instructions rather than failing silently.

Manual install requires Claude Code CLI and Python 3.10+; some features (PDF reports, landing page analysis) need optional dependencies (reportlab, Playwright). Audit quality depends on the data you provide (exports or screenshots). Garbage in equals generic benchmarks out. Some sub-skills (ads-amazon, ads-attribution, ads-server-side-tracking) are standalone in Wave 2; paired audit agents land in Wave 3. Results vary significantly by account maturity, budget level, and industry. Provide upfront context for relevant recommendations.