A new open-source project called GTM Engineer Skills packages nine Claude Code skills into a single pipeline for building websites that rank well in both traditional search engines and AI-powered search tools like ChatGPT, Perplexity, and Google's AI Overviews.
The toolkit, released under MIT license by onvoyage-ai, tackles a problem that's becoming more urgent by the month: optimizing content not just for Google's classic algorithm, but for generative engine optimization (GEO) - the practice of structuring content so AI search engines can find, parse, and cite it in their answers.
The Nine-Skill Pipeline
The skills run sequentially through a content production workflow:
- Brand DNA research pulls positioning, audience, and messaging from a company URL
- Keyword research identifies high-value terms using web search or paid tools like Ahrefs and Semrush
- GEO content research maps questions people ask AI engines about a product category, organized by intent (buying, problem-solving, learning)
- Content planning creates page blueprints with URLs, types, and priority levels
- Content writing generates markdown articles with metadata, with specific guardrails against fabricated statistics
- Chart creation produces machine-readable visualizations (SVG, HTML tables, JSON-LD) that AI engines can parse and cite
- Content auditing verifies facts and confirms URLs are live before publishing
- Page building converts audited markdown into production-ready frontend pages
- AEO/GEO optimization applies 16 checks across 6 dimensions to improve AI engine parsing
The project has picked up 428 GitHub stars since its February 2026 release, suggesting real interest in this niche. The skills are designed specifically for Claude Code, Anthropic's code execution environment, so you'll need that setup to use them.
Practical vs. Theoretical
What separates this from generic SEO advice is the emphasis on machine-readable formats. The chart creation skill, for example, outputs JSON-LD structured data alongside visual SVGs - giving AI engines clean data to reference rather than trying to interpret an image. The audit skill checks that sources are real and links are live, which matters more when AI engines are citing your content directly in their responses.
The GEO angle is still early. Nobody has definitive data on what makes content rank higher in AI search results, and the "16 foundational checks" the toolkit applies are based on emerging best practices rather than proven algorithms. But the research-first approach - mapping what questions AI engines get asked before writing anything - is a sound methodology regardless of how the optimization details shake out.
For teams already using Claude Code and producing content at volume, the pipeline structure alone could save hours of manual coordination between research, writing, and publishing steps.