The agent successfully assembled a comprehensive explanation of CockroachDB by working around the docs site's JavaScript-heavy rendering. The docs.cockroachlabs.com domain itself proved largely inaccessible to direct content extraction (serving only HTML scaffolding), forcing the agent to rely on web search results, the company's main marketing site, and prior knowledge to answer all four questions: what CockroachDB does, who it's for, how it's priced, and how it differs from alternatives. The site is poorly agent-ready for discovery and positioning work.

• ›Steps [2], [5], [6], [9], [10], [12] (web searches) were the primary content-bearing sources—they returned third-party summaries, comparison frameworks, and pricing breakdowns that the agent cited directly. The search results contained more actionable information than direct fetches from cockroachlabs.com domains.
• ›Steps [3] and [4] (main site and pricing page) were fetched but returned only HTML scaffolding/metadata; the actual pricing details and product description came from external sources like Airbyte's pricing explainer and DB-Engines, not from parsing the company's own pages.
• ›Step [8] (llms.txt) returned a machine-readable index of documentation links but no actual content—useful for site structure but not for answering the questions. The docs site itself (steps [1], [7], [11]) consistently returned client-side-rendered pages with minimal server-side HTML content, making them unreadable to the agent without JavaScript execution.
• ›The agent had to guess or use prior knowledge to navigate to /pricing, /docs/stable, and /frequently-asked-questions; these URLs were not discoverable from the homepage response, indicating poor information architecture for machine navigation.
• ›The final answer cites 12 sources, but only 3–4 are from cockroachlabs.com domains; the rest are external comparisons, third-party pricing analyses, and general knowledge. This heavy dependence on external sources reflects that the docs site failed to expose the required information in a fetchable form.