The agent successfully assembled a comprehensive explanation of Passkey Central by combining direct fetches of key resource pages with web search discovery. The site's Docusaurus-based structure made content accessible via search results and direct URLs, but the agent could not explore the full site architecture through the homepage alone—it relied heavily on search to discover specific resource pages and had to work around truncated HTML responses that prevented full content inspection.

• ›Content-bearing steps [10, 11, 12, 15, 16] all returned valid 200 responses to specific resource URLs (costs-and-benefits, design-guidelines, introduction-to-passkeys, passkey-roll-out-guides, developer-documents), which the agent discovered via web search rather than site navigation.
• ›The homepage and standard path guesses (/pricing, /about, /docs) all returned 404s or generic Docusaurus pages, forcing the agent to rely on web search (45% of fetches) to find actual content pages—indicating poor discoverability from the root domain.
• ›The site is content-rich but navigation-hostile: the agent could not construct a mental map from the homepage and had to search externally to locate the resource sections that answered questions about who it's for, what it does, and rollout strategies.
• ›The agent correctly identified a key insight—that Passkey Central is educational guidance, not a product—but this distinction was inferred from the resource structure and prior knowledge rather than explicitly stated in fetched content, suggesting the site could be clearer about its purpose upfront.