The agent successfully gathered comprehensive information about OpenPost by accessing markdown versions of the site's core pages (homepage, features, pricing, FAQ, platforms, and self-hosting). The site is highly agent-ready: it publishes structured markdown files alongside its JavaScript-heavy HTML, making key information fetchable and machine-readable. The agent synthesized this into a complete explanation of what OpenPost does, who it's for, how it's priced, and how it differs from alternatives.

• ›The agent relied entirely on markdown artifact versions (.md routes) for content extraction, bypassing the JavaScript-heavy HTML frontend. These markdown files were explicitly cited in the final response's sources section, indicating they were the primary content-bearing resources.
• ›The site deliberately publishes parallel markdown versions of all major pages (index.md, features.md, pricing.md, faq.md, platforms.md, self-hosted.md), which significantly improved agent navigability. This is an explicit accommodation for non-browser consumption.
• ›Pricing information was retrieved directly from /pricing.md and presented in structured tiers with clear per-plan limits (workspaces, accounts, posts/month, storage). The FAQ and features pages provided additional context about platform support, self-hosting, and feature availability guarantees.
• ›The agent identified genuine friction points: the site's careful distinction between 'Hosted service claims' (features with SLA guarantees) vs. features that work but lack guarantees; the JavaScript-dependency of the HTML frontend; and the lack of explicit competitor comparisons. These were accurately surfaced as confusing but did not prevent task completion.
• ›The agent did not need to execute actions (no auth, payment, or integration setup required) — this was purely a discovery-evaluation task requiring content extraction and synthesis, which the markdown layer supported well.