The agent assembled a functional explanation of NexusPool by combining direct site fetches with web search context, achieving a mostly complete answer about what the service does, who it's for, pricing, and differentiation. However, the site proved only partially agent-readable: the blog homepage returned truncated HTML, the main site is JavaScript-heavy and required prior knowledge to interpret, and the agent had to supplement with external sources (Hashrate Index) to validate and contextualize its findings. Key facts (0% fee, non-custodial model, Stratum V2, Glass Ledger) came from structured metadata and prior knowledge rather than clearly published prose.

• ›Step [1] (blog homepage) returned structured meta tags (title, description) that named key features (Stratum V2, solo mining odds, Proof of Custody) but the HTML was truncated, preventing full blog content access. The agent inferred service positioning from title alone.
• ›Step [3] (main nexuspool.io) is a Next.js SPA that returned only skeleton HTML with no rendered text content—the agent could not extract pricing, positioning, or feature details from the fetch itself and relied on prior knowledge instead.
• ›Step [7] (external Hashrate Index article) was the only fully readable source but was not NexusPool's own property. The agent cited it as a source yet it provided industry context rather than definitive NexusPool facts, indicating the primary domain did not self-publish comparison or positioning content.
• ›The agent correctly identified 0% fees and non-custodial operation but could not verify these directly from the fetched site content—they appear to come from prior knowledge or inferred from the name/metadata rather than explicit documentation on the domain.
• ›Site is low in agent-readiness for discovery-evaluation tasks: no plaintext blog posts were returned, no FAQ or pricing page was reachable, and JavaScript-rendered UI prevented content extraction. The agent had to fall back on web search and prior knowledge to answer the task.