/agents.md holds short, binding rules for an AI agent that integrates with Rivr: use the sandbox and only test identities, retry safely with an Idempotency-Key and the same order id, retry only when the error catalog says it helps, verify webhook signatures, never send a secret to the client, and…
This page needs JavaScript. The same content is available as Markdown: /changelog/2026-10-25-rules-for-ai-agents.md. Every document for agents is listed in /llms.txt.