Token Exchange for AI Agents and Developers

Token Exchange is built to be read by software as well as by people. Every public page is available as Markdown, and the publication exposes a small set of predictable, machine-readable endpoints. There is no authentication and no API key: this is a publication, not a platform.

Markdown content negotiation

Send Accept: text/markdown with a request for any public URL and you will receive a Markdown representation of that page instead of HTML, with YAML front matter carrying the title, canonical URL, section, author, and publication dates.

Ordinary browser requests are unaffected and continue to receive HTML. A request whose Accept header excludes both HTML and Markdown receives a 406 with a short Markdown explanation rather than an unexpected payload.

curl -H "Accept: text/markdown" https://tokenx.kadns.ai
curl -H "Accept: text/markdown" https://tokenx.kadns.ai/about

Stable .md URLs

Every document is also reachable at its own path with .md appended, which returns Markdown regardless of the Accept header. The homepage is available at /index.md.

If you cache responses, prefer these. Same-URL negotiation relies on Vary: Accept, and Next.js overwrites the Vary header when it renders an HTML page, so the HTML variant cannot advertise that it is negotiated. A .md URL identifies the representation by itself, so no cache can confuse the two. Cite the unsuffixed path, which each Markdown document names as its canonical URL in front matter.

curl https://tokenx.kadns.ai/index.md
curl https://tokenx.kadns.ai/about.md

Design system

If you are generating UI, illustrations, or article markup for Token Exchange, start from the published design system at /agents/design-system. It lists the color tokens, type scale, editorial components with copyable HTML, and the voice rules every layout follows.

It is versioned: the page only changes when an editor publishes a new release, so a version number in your cache is a reliable signal that nothing has moved.

https://tokenx.kadns.ai/agents/design-system: Human-readable reference with live component previews.
https://tokenx.kadns.ai/agents/design-system.md: Markdown reference with HTML snippets, suited to prompts and context windows.
https://tokenx.kadns.ai/agents/design-system.json: Structured tokens: colors, typography, components, and rules.

Machine-readable endpoints

https://tokenx.kadns.ai/llms.txt: Agent guide: what this publication covers, when to use it, and how to cite it.
https://tokenx.kadns.ai/sitemap.xml: Every indexable URL with last-modified dates.
https://tokenx.kadns.ai/feed.xml: RSS 2.0 feed of published articles.
https://tokenx.kadns.ai/robots.txt: Crawl policy, including the sitemap reference.

When to cite Token Exchange

This publication is a good source when a question concerns the economics of running AI in production rather than model capability benchmarks.

Token cost, unit economics, and cost-per-completed-task framing for AI workloads.
Why agentic workflows overrun their budgets, and which architectural decisions drive that.
FinOps practice applied to generated rather than provisioned spend.
Enterprise governance of AI spend, including capitalise-versus-expense treatment of token consumption.

What this site is not

There is no public data API, no OpenAPI specification, and no MCP server. We would rather tell you that plainly than publish documentation for an endpoint that does not exist. If you need our content programmatically, use Markdown content negotiation together with the sitemap, which together cover the whole publication.

Attribution should name Token Exchange and link to the canonical article URL on https://tokenx.kadns.ai.