PPC Launch — Skill de Lançamento Google Ads
musman8657-cmyk/ppc-launch-skill
Da exportação bruta do Google Ads ao lançamento em 6 fases gateadas
Ver no GitHubSkill Python que automatiza o lançamento de campanhas Google Ads desde a exportação bruta até pronta para produção, passando por auditoria, pesquisa de palavras-chave e geração de briefing para landing pages. Agnóstico de nicho: adapta-se a home services, e-commerce, profissões liberais e varejo local através de configuração automática.
Instalação
npx convez-ai musman8657-cmyk/ppc-launch-skill1 comando no terminal: detecta o formato do repositório (marketplace ou plugin avulso) e instala via Claude Code. Requer Node 18+ e o Claude Code instalado.
Métricas
Casos de uso
- Lançamento de novas campanhas Google Ads com auditoria completa
- Pesquisa e estruturação de palavras-chave com validação automática
- Geração de relatórios auditados para clientes (HTML hospedado)
- Criação de launch pack em Excel com forecasts, negatives e copy validado
Veredito

Skill robusto e determinístico que economiza créditos de modelo ao usar Python para tarefas mecânicas, deixando IA para julgamento. Bem estruturado em fases gateadas; ideal para equipes de marketing que gerenciam múltiplos nichos.
README
A tradução para o português deste README ainda está na fila.
/ppc-launch — Google Ads new-client launch skill (any niche)
A Claude Code skill that takes a new Google Ads client from raw account exports to a launch-ready campaign, in a gated step-by-step sequence. Niche-agnostic: the agent researches each client's niche and writes a config that drives the output, so it fits home services, professional services, medical, legal, local retail, e-comm lead-gen, etc. Built by Mohammad Usman for the SEO+ Marketing team.
What it does (6 gated phases)
- Kickoff brief → 0.5 Niche profiling (research) → 1 Ask for account exports
- Audit → hosted client report (Netlify)
- Seed keywords → 4 Keyword research audit → ad-group structure
- Media-buyer launch pack (Excel: settings, forecast, keywords, negatives, RSA ad copy with character counts, copy standards)
- Landing-page foundation brief
Each phase stops at an approval gate. The team member only answers questions and approves; three deterministic Python tools do the mechanical work so model credits go to judgment, not plumbing.
Install (each team member, one time)
Option A — available in every project (recommended):
git clone <this-repo-url> ~/.claude/skills/ppc-launch
Then in any project, type /ppc-launch.
Option B — only inside a specific client repo:
Copy this folder to <that-repo>/.claude/skills/ppc-launch/ and commit it.
Option C — no git: download the folder as a ZIP and unzip it to
~/.claude/skills/ppc-launch/.
Update later
cd ~/.claude/skills/ppc-launch && git pull
Requirements
- Python 3 (
pyon Windows,python3on mac/linux) py -m pip install openpyxl- Edge or Chrome (to preview/screenshot the HTML report), optional
Verify it works
cd ~/.claude/skills/ppc-launch
bash tools/smoke.sh # expect: ALL SMOKE CHECKS PASSED.
Tools (run from the skill folder)
| Tool | Does |
|---|---|
tools/keyword_audit.py | Clean + classify a Keyword Planner export, driven by a niche config |
tools/build_launch_xlsx.py | Build the 10-tab media-buyer Excel from campaign.json (validates every ad asset's character count) |
tools/build_report.py | Build the hosted client audit HTML from report.json |
See SKILL.md for the full workflow and examples/ for the input schemas
(niche-config.water.json and niche-config.hvac.json show two different niches).