GEO-SEO Claude goes viral as AI search visibility reshapes SEO agencies

A free Claude-powered auditor for AI search visibility has become an uncomfortable stress test for SEO agencies: the diagnostic layer is getting cheaper, while the work clients still struggle to buy is moving toward strategy, implementation, experimentation, and executive interpretation.

Zubair Trabzada’s geo-seo-claude repository sat at about 6.4K stars and about 1K forks on GitHub, with the project described as a GEO-first SEO skill for Claude Code. The README calls it a GEO-first, SEO-supported tool and says it is MIT licensed. Inside the repository, the workflow goes well beyond a basic crawl. It is built to analyze citability scoring, AI crawler access, llms.txt compliance, brand mention presence across AI-cited platforms, technical infrastructure, content quality, schema markup, and PDF reporting. That is exactly why it is landing so hard: a slice of what agencies once wrapped into a monthly retainer is now packaged as code.

The timing matters. GEO, short for Generative Engine Optimization, entered the conversation in a 2023 paper that framed it as a way to improve visibility inside generative engine responses. Google then tested AI Overviews in main search results on March 22, 2024, and officially launched them in the United States on May 14, 2024. The old blue-link model did not disappear, but it stopped being the only game in town. As AI summaries and answer boxes take more real estate, the fight has shifted from ranking alone to being cited, summarized, and retrieved by the systems users are actually reading.

The traffic numbers show why the panic around agency pricing has intensified. Search Engine Land reported AI-sourced sessions across 19 GA4 properties rose from 17,076 between January and May 2024 to 107,100 in the same period in 2025, a 527% increase. Semrush later found AI Overviews appearing for 30% of U.S. desktop keywords as of September 2025. That is the kind of shift that turns a niche acronym into a boardroom problem.

Claude itself is part of the story. Anthropic says Claude Code can read codebases, make changes across files, run tests, and deliver committed code, while its Skills system packages repeatable workflows and domain expertise for Claude. Anthropic also released updated Sonnet and Opus models in 2026 with stronger coding and agentic performance. Put together, the message to agencies is blunt: if a free open-source tool can audit visibility, the retainer has to justify itself somewhere else. The future belongs to the firms that can turn the audit into action, then prove what changed.