Claude Ads
A comprehensive paid advertising audit and optimization skill for Claude Code covering Google, Meta, YouTube, LinkedIn, TikTok, Microsoft, and Apple Ads with 250+ audit checks.
At a Glance
About Claude Ads
Claude Ads is an open-source paid advertising audit skill for Claude Code that delivers 250+ structured checks across seven major ad platforms. It runs entirely on your local machine, analyzing data you provide — exports, screenshots, or pasted metrics — without connecting to any ad platform automatically. The tool spawns parallel subagents for full multi-platform audits, generates a weighted Ads Health Score (0–100), and produces professional PDF reports for client deliverables.
- 250+ Audit Checks across Google Ads (80), Meta Ads (50), Apple Ads (35+), TikTok Ads (28), LinkedIn Ads (27), Microsoft Ads (24), and cross-platform checks — each with weighted severity scoring.
- Parallel Subagent Delegation via
/ads auditspawns 6 simultaneous audit agents (Google, Meta, creative, tracking, budget, compliance) for maximum speed. - Ads Health Score (0–100) with letter grades (A–F) and a prioritized action plan generated after every full audit.
- Industry-Specific Templates for 11 business types (SaaS, ecommerce, local service, B2B enterprise, healthcare, finance, and more) loaded via
/ads plan <type>. - PPC Financial Modeling with
/ads mathfor CPA, ROAS, break-even, budget forecasting, and LTV:CAC calculations. - A/B Test Design via
/ads testcovering hypothesis frameworks, statistical significance, sample size, and duration planning. - PDF Report Generation via
/ads reportwith health score gauges, platform comparison charts, and pass/fail distributions for client deliverables. - Quality Gates enforce hard rules on every audit: Broad Match without Smart Bidding, 3x Kill Rule for CPA, budget sufficiency thresholds, learning phase protection, and privacy infrastructure checks (Consent Mode V2, CAPI, Events API, AdAttributionKit).
- Optional Live Data Integration via MCP servers (mcp-google-ads, Adspirer, GrowthSpree) for direct API access without manual exports.
- 25 Reference Files with 2026-current benchmarks, bidding decision trees, platform specs, and compliance requirements loaded on demand using a RAG pattern.
- Privacy-First Architecture — all analysis runs locally; no ad account data is sent to external servers.
- Get started by installing via Claude Code plugin marketplace (
/plugin marketplace add AgriciDaniel/claude-ads) or with a one-command shell installer, then run/ads auditto begin.
Community Discussions
Be the first to start a conversation about Claude Ads
Share your experience with Claude Ads, ask questions, or help others learn from your insights.
Pricing
Open Source (MIT)
Fully free and open-source under the MIT License. Use, modify, and distribute freely.
- 250+ audit checks across 7 ad platforms
- Parallel subagent delegation
- Ads Health Score (0-100)
- Industry-specific plan templates
- PPC financial modeling
Capabilities
Key Features
- 250+ audit checks across 7 ad platforms
- Parallel subagent delegation for full audits
- Ads Health Score (0-100) with letter grades
- Industry-specific ad plan templates (11 business types)
- PPC financial modeling (CPA, ROAS, break-even, LTV:CAC)
- A/B test design with significance and sample size
- PDF audit report generation for client deliverables
- Quality gates and hard rule enforcement
- Creative pipeline with 4 specialized AI agents
- Optional MCP server integration for live API access
- 25 RAG reference files with 2026-current benchmarks
- Local-only data processing for privacy
- Auto industry detection from ad account signals
- Cross-platform creative quality audit and fatigue detection
- Competitor ad intelligence across all platforms