The agent could not access eon.com directly due to Cloudflare protection, but assembled a comprehensive explanation of E.ON's business, pricing, and market position entirely from web search results and third-party sources (comparison sites, Wikipedia, news articles, regulatory filings). The site itself was not navigable; the agent worked around complete inaccessibility by relying on 80% web search and 20% prior knowledge to fulfill the discovery-evaluation task.

• ›Step [3] search results included Wikipedia and comparison-site URLs that provided core business structure information; steps [8], [9], [10], [11] returned pricing details, competitor comparisons, and regulatory/structural insights from third-party aggregators and news sites.
• ›E.ON's own website (eon.com) published no machine-readable, agent-accessible content — the Cloudflare protection blocked all direct fetches [1, 4, 5, 6, 12], forcing the agent to reconstruct the answer from external sources rather than official documentation.
• ›The site's agent-readiness is extremely low: no structured data (schema.org), no llms.txt, no API, and aggressive bot-blocking via Cloudflare. The agent had to assemble fragmented information from Wikipedia, price-comparison sites, and news outlets to build a coherent narrative about services, pricing tiers, and competitive positioning.
• ›Agent cited web search results in its final Sources section but those citations refer to third-party sites (MoneySuperMarket, Which?, Wikipedia, Investing.com), not to eon.com itself — indicating the primary source domain was entirely inaccessible and the explanation is indirect.
• ›Gaps in the final response (unclear business-segment interaction, pricing complexity, recent restructuring context) reflect the limits of secondary-source inference when primary source material is blocked.