Stripe's Agent-Readiness Gap
Public-surface analysis of stripe.com. Surfaces checked: HTTP response headers, robots.txt, sitemap files, visible HTML + JSON-LD on /, /pricing, /payments, /connect, llms.txt, and /.well-known/ endpoints. No auth, no internal tooling, no proprietary data.
// editorial note: findings marked TODO require live-crawl verification before final publication.
TODO: verify against live crawl. Initial review suggests /payments lacks Service or SoftwareApplication JSON-LD describing the core Payments product. Confirm before publishing.
TODO: verify. Pricing pages for complex products often ship a single Offer schema entry that doesn't reflect the full pricing matrix. Stripe's interchange-plus model is a likely candidate.
TODO: verify whether /.well-known/agents.json or an MCP server registry is present. If absent, this is a high-priority gap given Stripe's agent integrations.
TODO: verify /llms.txt presence and quality. Stripe's docs are structured well for crawlers; an llms.txt with anchors for /payments, /connect, /billing, /terminal would consolidate that surface.
TODO: verify presence of OAuth metadata at well-known endpoints for Stripe Connect. Scope documentation is strong; discoverability of the metadata endpoint should be confirmed.
Stripe publishes an OpenAPI specification at a stable URL and links it from developer documentation. Agent retrieval pipelines have a clear machine-readable entry point for the API surface.
TODO: confirm. Stripe's robots.txt is expected to allow major AI crawlers (GPTBot, ClaudeBot, PerplexityBot). Verify Content-Signal directives if present.
Stripe's API returns structured JSON errors with machine-readable codes (`type`, `code`, `param`, `decline_code`). This is a reference implementation for agent-tolerant error semantics.
The split between developer and marketing surface
Stripe's developer surface (API, docs, error semantics) is among the best-documented in the industry. The marketing surface - product pages, pricing, MCP discoverability - appears to lag. This is a common pattern: teams who built the API for machines decades ago did not extend that discipline to the pages designed for humans. The gap matters now because agents read both.
Methodology note: all findings are based on public HTTP responses as of 2026-06-10. Items marked TODO require live-crawl verification before final publication. No claims are made about internal systems, analytics, revenue, or implementation quality beyond what is directly observable in the crawl output.
See if your site has these gaps.
The Agent-Readiness Gap Audit applies the same six-surface methodology to your product. Scorecard, surface map, prioritized remediation tickets.
Get the next teardown
Technical teardown, no fluff. One email when the next agent-readiness teardown goes live.