The agent successfully assembled a complete explanation of Robotomail's service, pricing, positioning, and differences from alternatives by fetching markdown versions of the homepage, docs, and skill pages. The site is agent-friendly in that it publishes structured markdown alongside JavaScript-rendered HTML, allowing the agent to extract detailed information without needing to parse rendered DOM; however, pricing details were only clearly visible in markdown form, not in the rendered pricing page itself.

• ›Step [6] (index.md) provided the core value proposition and feature overview in markdown; step [7] (docs.md) delivered technical details, API structure, and inbound/outbound mechanics; step [11] (skill.md) gave agent-specific self-onboarding instructions. Together these formed a complete mental model.
• ›The agent discovered markdown endpoints by inference after encountering JavaScript-rendered HTML pages [1, 3, 4, 9]—it recognized the pattern and proactively requested .md versions rather than being explicitly directed. This suggests the site does not advertise markdown availability in navigation or meta tags, forcing agent-discovery rather than surfacing it natively.
• ›Pricing tiers and feature comparison are embedded in markdown documentation [6] rather than exposed as structured data (JSON schema, microdata, or a dedicated machine-readable pricing endpoint), requiring the agent to parse natural language. The site's agent-readiness for pricing discovery is moderate: markdown is better than DOM parsing, but worse than JSON or schema.org pricing.
• ›The agent identified several gaps (rate limit specifics, webhook retry policy, HMAC signing format, storage quota behavior) that were either undocumented or only partially documented in the accessible markdown, suggesting some technical depth is missing from public docs.