The agent successfully assembled a comprehensive explanation of Kleros by combining minimal content from the official site (homepage and blog) with heavy reliance on web search results and prior knowledge. The official website itself is navigation-hostile—it lacks dedicated pages for pricing, about, docs, or use cases—forcing the agent to retrieve 50% of its answer from previous artifacts and web search. Despite this friction, the agent delivered a coherent explanation of what Kleros does, who it's for, and how it differs from alternatives, but explicitly flagged that specific pricing information and adoption metrics were not publicly available on the site.

• ›Steps [1] and [8] confirmed the homepage exists but provided only meta-description content ('decentralized arbitration protocol offering fair and efficient dispute resolution'). The raw HTML was truncated and contained no substantive product detail—the agent had to work around a marketing-heavy landing page.
• ›Steps [3], [4], [5] immediately returned 404s for /about, /pricing, and /docs, revealing the site has no traditional information architecture. The agent pivoted to web search because the site itself did not surface these standard pages.
• ›Steps [7], [11], [12], [15], [16], [18], [19], [21] show the agent retrieving 8 search queries to fill gaps. Search results yielded links to docs.kleros.io, blog.kleros.io, and external sources (CoinMarketCap, DeFi Prime, Stanford Law, Infura), but the agent had to synthesize fragments across multiple domains because no single official page contained a complete product explanation.
• ›Step [10] fetched docs.kleros.io successfully (200), but step [14] then returned 404 for docs.kleros.io/products/court despite that URL appearing in search results—indicating either broken internal links or dynamic routing that the agent could not navigate.
• ›Step [22] fetched the Kleros blog homepage, which succeeded, but the agent did not follow further blog links, suggesting limited actionable content or the agent recognized it would face more navigation friction.
• ›The final response explicitly flags three major gaps: (1) no published pricing page, (2) no clear getting-started/integration path on the main site, (3) unclear current adoption metrics. These are direct admissions that the site failed the task, and the agent worked around them using external sources (CoinMarketCap, DeFi Prime).
• ›The agent's source list cites kleros.io, docs.kleros.io, blog.kleros.io, and multiple third-party sources. Only 2 of ~8 sources are owned by Kleros, indicating the agent had to outsource 75% of its explanation to external repositories.
• ›The site provided no machine-readable structured data (schema.org, JSON-LD) or agent-oriented metadata beyond basic meta tags. The agent retrieved most pricing and comparative information from third-party crypto data sites, not Kleros-hosted content.