AI-readable corpus

This documentation site is written for both humans and AI agents. Machine consumers get two files, generated at build time from every page's frontmatter and body:

File	Contents
/llms.txt	Curated index: page → description → URL. Point an agent here first to choose the pages it needs.
/llms-full.txt	The full corpus, concatenated in reading order with page separators. For agents that should read everything.

Intended use

• Start at /llms.txt. Each line lists one page, its one-line description, and its URL. Select pages relevant to the task, then fetch them (plain .md is available at the same path for every page, e.g. /api/payments.md).
• For deep tasks (implementing a full integration), fetch /llms-full.txt once and process the whole corpus — it is ordered so that concepts precede the reference pages that build on them.
• For agent self-integration, the MCP and A2A pages contain copy-paste client configuration so an agent can connect from the documentation alone.

Page index

The authoritative index is generated into /llms.txt at build time from the page frontmatter below; the list is maintained there and on this site's Overview. The site's reading order:

1. Overview
2. Quickstart
3. Sandbox
4. Concepts: Payments · Tokens · Accounts · Plans
5. Checkout: Snippet · Secure Fields
6. API: Overview · Authentication · Payments · Refunds · Webhooks · Errors · Rate limits
7. Agents: MCP · A2A
8. Operations: Go live · SCA · Disputes · Settlement · Security · PCI posture
9. Payment flow & roles

Raw markdown mirrors

Every page is also available as raw markdown at its URL with a .md extension (/concepts/payments.md, /api/webhooks.md, …) — convenient for agents and for diffing against a pinned version of the docs.