Claude SEO's 19-Agent Architecture: How 26 Sub-Skills Coordinate Technical Audits Without a Central Orchestrator

Claude SEO utilizes a 19-agent architecture composed of 26 sub-skills to conduct thorough technical SEO audits. The system lacks a central orchestrator, relying instead on a flat skill composition pattern to coordinate tasks. Each sub-skill wraps one or more agents and exposes a uniform interface to Claude Code's skill system. This design offers benefits such as simpler failure modes, easier agent reuse, and clearer boundaries, but also results in the loss of explicit dependency graphs, centralized state management, and the ability to optimize cross-agent workflows.

The system defines 26 sub-skills that wrap 19 specialist agents covering various aspects of technical SEO. These include Technical SEO agents for crawl analysis, robots.txt validation, sitemap parsing, canonical tag checks, and redirect chain detection. Performance agents measure Core Web Vitals, detect render-blocking resources, and analyze JSON-LD parsing for schema markup.

Schema agents validate structured data, parse JSON-LD, and assess rich snippet eligibility. E-E-A-T (expertise, authoritativeness, trustworthiness) agents extract author bios, perform citation analysis, and detect expertise signals. AI search agents ensure GEO/AEO readiness, generate llms.txt files, and support WebMCP integration. Domain-specific agents handle local SEO, e-commerce, international SEO, and backlink analysis.

Each skill exposes a standard interface with methods for analysis, dependency declaration, and context merging. Claude Code invokes skills in dependency order, with skills declaring dependencies explicitly. The system resolves a directed acyclic graph of dependencies and executes skills in topological order. A single AuditContext object is shared among all skills, enabling them to read from and write to a common source without message passing or event buses.

This shared-memory model creates predictable failure modes, such as missing dependencies, context pollution, and concurrent access issues.

The composition pattern allows reuse of agents across skills and offers different composition types, such as sequential pipelines, parallel merges, sequential fallbacks, parallel aggregations, and independent checks. Optional live data integrations can be added through a nine-extension system, including integrations with various SEO platforms like Google Search Console, Google Analytics, Google PageSpeed Insights, Ahrefs, SEMrush, Moz, and Screaming Frog.