For machines and the people who write them

Developers & AI agents.

Everything on this page describes content that is already public. Four read-only JSON documents, an OpenAPI description, and a plain-Markdown copy of every published article. No API key, no rate limit, no write path.

Start here

Every page on this site carries Link headers pointing at the documents below, so an agent that lands anywhere can find them without being told.

The four endpoints

All four are static JSON, rebuilt on every deploy and served from the edge. GET only. CORS is open.

Article index

/api/article-index.json

One row per published English article — slug, silo, editorial template and title. No body text; fetch the page or its markdown sidecar for that.

{ summary, articles[] }

Navigation manifest

/api/navigation-manifest.json

The mega-menu as data: every surface, the links inside it, the slot caps the renderer enforces and the guardrails that check them.

{ brand, generated_at, limits, surfaces, files, guardrails }

Reverse-silo targets

/api/reverse-silo-targets.json

For each insurance silo, the hub page and the action page that articles in that silo link toward.

{ summary, targets[] }

Spanish article inventory

/api/es-manifest.json

The /es/ locale, which has no markdown sidecars — Spanish articles are stored as JSON, not Markdown. Slug, silo, action-page target and internal-link state.

{ summary, articles[] }

Markdown, not scraped HTML

Every published English article has a plain-Markdown twin at /md/{slug}.md — the same body the page renders, without the navigation, widgets or CTAs you would otherwise have to strip out.

https://www.simplyinsurance.com/best-life-insurance-companies/ https://www.simplyinsurance.com/md/best-life-insurance-companies.md

Drafts and unapproved articles have no sidecar — if a URL renders a page but /md/ returns 404, that is the gate working, not a bug. Spanish articles have no sidecars at all; use the /api/es-manifest.json inventory instead.

How you may use it

Our robots.txt carries Content Signals on every group. Access and usage are separate questions; these are the usage half.

  • search yes Index this content and link to it from results.
  • ai-input yes Retrieve it at answer time to ground a generated response. This is the one that produces a citation, and a citation is what sends us a reader.
  • ai-train no Do not retain it in a training corpus.

Why ai-train=no: over one representative day, one provider sent us 11,990 crawl requests and zero answer-engine referrals, while another sent 4,460 and returned 4,713 answer requests. Grounding and citing earns the fetch. Ingesting without sending a reader is a request for the corpus, not a readership. Figures from our Cloudflare AI Crawl Control dashboard, 24 hours to 19 August 2026.

What is deliberately not here

Agents infer capability from documentation, so these absences are stated rather than left to be discovered:

  • No transaction endpoint. There is no quote, checkout, payment or lead-submission API on this host. Quotes are handed off to licensed carrier partners through the action pages listed in /llms.txt, and those handoffs are browser flows. Do not synthesise one from the read endpoints above.
  • No authentication. Nothing here is private, so there is no OAuth server, no token endpoint and no /.well-known authorization metadata. Anything claiming otherwise for this domain is not us.
  • No agent-to-agent service yet. If and when one ships it will announce itself from an agent card in /.well-known/. There isn't one today.

Content is geo-restricted to the United States, US territories and Canada. Verified search and AI crawlers are allowlisted; a fetch from elsewhere may be blocked at the edge regardless of user-agent. Questions: support@simplyinsurance.com.