Ivan Falco's Ads Skills for Claude Code

Run your B2B paid motion like an ads engineer. 40+ strategy files, 39 API scripts, and battle-tested frameworks for LinkedIn, Meta, and Google Ads - built from managing $200K+/month in B2B ad spend across 12+ accounts.

Clone this repo, open Claude Code, and ask it anything about your ads. It knows the strategy, it pulls your live data, and it manages your campaigns.

Important: this runs in Claude Code on your own computer - the claude CLI in your terminal, or the Claude Code desktop app (Local mode). It does not work in the Claude chat app (claude.ai / the Claude desktop chat), which can't run scripts on your machine. New to Claude Code? Install it: https://code.claude.com/docs

Built by

Ivan Falco - I do ads engineering: I help B2B companies scale their paid motion using smart AI systems - ABM and ABM 1:1, syncing ad audiences with outbound, AI-native creative, and scaling accounts like an engineer, at Frontal (formerly ColdIQ Agency).

Want a human to look at your setup? Connect with me on LinkedIn and send a connection request with a note: https://www.linkedin.com/in/ivanfalco/

---

What's Inside

4 Skills

40+ Knowledge Base Files

Ads Foundations (11 files):

• 5-Stage Demand Engine (replaces traditional TOFU/MOFU/BOFU)
• Budget allocation by stage and channel
• Ad copywriting frameworks (voice-of-customer, 5-layer audit)
• Writing style guide (no AI slop - governs all copy and replies)
• Channel selection criteria
• Optimization signals (leading vs lagging)
• Scaling quadrant framework
• Offers strategy by funnel stage

LinkedIn Ads (15 files):

• Full-funnel framework (TOF/MOF/BOF with budget splits)
• Audience sizing rules (60K-400K cold, retargeting ranges)
• 6 campaign structure models + naming conventions
• Bidding strategy (automated → manual CPC progression)
• Creative by awareness stage (12 angles with rationale)
• ABM playbook (1:1, 1:few, 1:many approaches)
• Scaling progression (penetration-based)
• Format-specific: conversation ads, document ads, CTV
• 35-item audit checklist
• CTR/CPC/CPL benchmarks by funnel stage

Meta Ads (16 files):

• Meta Ads Operating System (the master decision framework)
• Creative Cadence OS (production pipeline, iteration hierarchy)
• Why Meta works for B2B (50% lower CPL vs LinkedIn when done right)
• Pixel + CAPI setup and event hierarchy
• Audience strategy (CRM lookalikes → third-party → broad)
• 3-phase campaign structure (ABO → CBO → Advantage+)
• Creative fatigue detection and rotation cadence
• Advantage+ automation guide
• Optimization playbook with B2B benchmarks
• Lead form optimization (social amnesia problem)
• ABM on Meta playbook
• Offer strategy by funnel stage

Google Ads (9 files):

• Intent-first strategy (the intent ladder - capture demand before creating it)
• Account structure (split by intent, themed ad groups, naming, default settings to fix)
• Keywords and match types (the Phrase/Exact -> Broad progression, negatives)
• Bidding strategy (by conversion volume, tCPA/tROAS, the optimize-to-quality trap)
• Search terms and negatives (the weekly ritual, what to cut)
• Benchmarks and measurement (B2B ranges, offline conversion import, weekly scorecard)
• Performance Max for B2B (when to run it, guardrails, how to read it)
• RSAs and landing pages (asset strategy, pinning, message match, Quality Score)
• Campaign types cheatsheet (Search, Shopping, Display, PMax, Video)

39 Python Scripts

LinkedIn Ads - 14 scripts:

• account_overview.py - Account dashboard with period comparison
• list_campaigns.py - All campaigns with status, budget, metrics
• get_campaign_performance.py - Detailed analytics with daily breakdown
• create_campaign.py - Create campaigns (5 objectives, 3 bid strategies)
• update_campaign.py - Update status, budget, bids, name
• list_creatives.py - All creatives with type and campaign association
• get_demographics.py - 5-pivot demographics (job function, seniority, company size, industry, country)
• upload_audience.py - Upload TAL/contact lists as DMP segments
• list_lead_forms.py - All lead gen forms with questions
• manage_bids.py - View and update bid strategy and amounts
• linkedin_api.py - Core API client class
• oauth_server.py - OAuth token flow
• config.py / client.py - Shared configuration and auth

Meta Ads - 12 scripts:

• account_overview.py - Account dashboard with actions breakdown
• list_campaigns.py - Campaigns with inline insights
• get_campaign_performance.py - Analytics with daily breakdown
• create_campaign.py - Create campaigns (5 objectives, special ad categories)
• update_campaign.py - Update status, budget, name
• list_ad_sets.py - Ad sets with targeting summary and metrics
• list_ads.py - All ads with performance data
• get_active_ads_copy.py - Full creative/copy extraction (link, video, carousel, dynamic)
• create_custom_audience.py - Upload hashed customer lists
• ad_scheduler.py - Schedule automatic ad pauses
• config.py / client.py - Shared configuration and auth

Google Ads - 13 scripts:

• account_overview.py - Account snapshot with period comparison
• list_campaigns.py - All campaigns with metrics
• get_campaign_performance.py - Detailed analytics with daily/custom ranges
• create_campaign.py - Create campaigns (6 types, 5 bidding strategies)
• update_campaign.py - Update status, budget, name
• create_ad_group.py - Create ad groups with CPC bids
• create_ad.py - Create RSAs with headline/description validation
• list_ads.py - Ads with performance and approval status
• add_keywords.py - Add positive/negative keywords (broad, phrase, exact)
• get_keyword_performance.py - Keyword analytics with Quality Score
• search_terms_report.py - Search terms audit, wasted spend finder
• config.py / client.py - Shared configuration and auth

Quick Start

# 1. Clone this repo
git clone https://github.com/ivangfalco/ads-skills.git
cd ads-skills

# 2. Open Claude Code
claude

# 3. Run the onboarding (5 minutes)
/onboarding

Or skip onboarding and just start asking:

"Audit my LinkedIn ad account"
"What's the right budget split for $50K/month?"
"Pull my active Meta ads and review the copy"
"Create a search campaign for our product"
"Which campaigns should I kill and which should I scale?"

How It Works

These skills turn Claude Code into a specialized advertising assistant. Each skill file teaches Claude:

• What to do - strategy frameworks, decision trees, benchmarks
• How to do it - Python scripts that connect to ad platform APIs
• When to reference what - routing logic that loads the right knowledge for the task

The methodology is the Ivan Falco 5-Stage Demand Engine approach - battle-tested across 12+ B2B accounts, $200K+/month in managed spend. When Claude gives you advice through these skills, it's grounded in real campaign data, not generic best practices.

Repo Structure

ads-skills/
├── CLAUDE.md                           # AI context - branding, rules, architecture
├── .claude/
│   └── skills/
│       ├── onboarding/                 # Interactive setup (SKILL.md)
│       ├── linkedin-ads/               # LinkedIn Ads skill
│       │   ├── SKILL.md                # Routing logic + methodology
│       │   ├── api-reference.md        # LinkedIn Marketing API docs
│       │   ├── knowledge-base/         # 15 strategy files
│       │   └── scripts/               # 14 Python scripts
│       ├── meta-ads/                   # Meta Ads skill
│       │   ├── SKILL.md
│       │   ├── api-reference.md
│       │   ├── knowledge-base/         # 16 strategy files
│       │   └── scripts/               # 12 Python scripts
│       └── google-ads/                 # Google Ads skill
│           ├── SKILL.md
│           ├── api-reference.md
│           ├── knowledge-base/         # 9 strategy files
│           └── scripts/               # 13 Python scripts
├── ads-foundations/                     # 10 cross-platform advertising frameworks
├── .env.example                        # Credential template
└── README.md                           # This file

Requirements

• Claude Code installed
• Python 3.10+ (for running API scripts)
• API credentials for the platforms you use

Who This Is For

• B2B marketers managing ad campaigns across multiple platforms
• Growth operators who want AI-assisted campaign management
• Agencies managing multiple client accounts
• Founders running their own paid acquisition

If you want to see what a full AI-native advertising operation looks like, check out what we're building at Frontal (formerly ColdIQ Agency).

License

Source-available: MIT + Commons Clause - see LICENSE. Use it, fork it, build on it, run it for your own and your clients' accounts. You just can't repackage and resell the skills themselves as a product. Attribution appreciated.

---

Built by Ivan Falco at Frontal (formerly ColdIQ Agency). Provided as-is. You are responsible for your own API usage, ad spend, and platform compliance.