ora
kernel.sh
success
Claude Code · Haiku 4.51:07
intent

What does kernel.sh do and who is it for? Explain it back to me.

13steps
67.6sduration
$0.8358cost
139,509tokens
22 steps9 reasoning steps6 searches
home
docs
docs
search
docs
docs
search
docs
docs
search
search
search
search
43%
on-site discovery
100%
reliability
29%
link following
path origin
  • previous resource29%
  • web search57%
  • prior knowledge14%
insight

The agent successfully compiled a comprehensive explanation of Kernel.sh by assembling information from the site's homepage, pricing page, and docs, supplemented with third-party comparison articles and blog posts. The site itself provided core positioning and pricing structure, but lacked a unified, machine-readable feature matrix and transparent unit-to-cost mapping, forcing the agent to rely heavily on external sources (Skyvern blog comparisons, search snippets) to answer the differentiation and detailed pricing questions completely.

  • Steps [1], [3], [4] (direct site navigation) provided foundational positioning and links to docs/pricing, but the HTML responses were truncated in the trajectory and no readable text was visible—the agent likely had to rely on prior knowledge of what those pages contain rather than parsing fresh content.
  • Steps [6], [11], [17], [18], [20] (web searches) were critical to answering differentiation vs. Browserbase and pricing details; search result titles and snippets contained the comparative information (speed benchmarks, pricing tiers, architecture differences) that the main kernel.sh site did not clearly surface.
  • Steps [14] and [15] (Skyvern blog articles on Kernel reviews and Browserbase vs. Kernel comparison) were explicitly cited and were the primary sources for the detailed feature comparison table and pricing breakdown—this third-party content was more comprehensive and agent-readable than kernel.sh's own marketing pages.
  • The site fragments information across multiple domains (/pricing, /docs, /blog) without a unified, structured presentation; the agent had to piece together understanding from navigation hints and external comparisons rather than finding a single authoritative resource.
  • Pricing structure (units, GB-seconds cost, tier inclusions) and open-source scope were acknowledged in the final response as confusing or missing—indicating the site's docs did not clearly expose these details in machine-readable form.

Want to run your own?

Join the waitlist for early access to point your own agents at any domain, with the intents you choose.

or talk to us about agent readiness →