The agent successfully assembled a comprehensive explanation of Supabase by relying primarily on web search results and third-party pricing guides rather than the site's own rendered content. The site's React-based architecture made direct HTML fetches unhelpful, forcing the agent to supplement with external sources; however, it did locate and cite Supabase's own Firebase comparison page, demonstrating that key marketing content exists on-site even if not easily machine-readable.

• ›Steps [1], [2], [3], [5] fetched Supabase pages directly but returned only HTML skeletons with no readable text content—the site is a React SPA and critical information is client-rendered, making it opaque to initial page fetches.
• ›Steps [7] and [8] (web searches) yielded third-party pricing guides and comparison articles that contained concrete, structured information (pricing tiers, compute costs, MAU limits, feature comparisons) that the agent extracted directly into the final answer.
• ›Step [9] successfully fetched Supabase's own Firebase comparison page (supabase.com/alternatives/supabase-vs-firebase), which was discovered via search and cited in the final response, proving that differentiation content exists on-site but is not discoverable through static HTML inspection.
• ›The agent explicitly flagged that the site's React rendering made direct content extraction impossible, and compensated by building the answer from third-party sources—a workaround that succeeded but reveals the site is not agent-ready for raw fetch inspection.
• ›Pricing details (four tiers, compute add-ons, MAU overages, spend caps) came entirely from external guides, not from the site's own /pricing page fetch; the agent had to rely on prior knowledge (50%) and web search (17%) rather than freshly crawled site content.