# A skill for Claude Code: the rules and the way in for AI agents

- Date: 2026-10-25
- Type: New
- Surface: Developer portal and CLI
- What you need to do: Nothing.
- Address: https://developers.rivr.io/changelog/2026-10-25-agent-skill-for-claude-code

The portal now serves a skill for Claude Code: /skills/rivr/SKILL.md, with three reference files under /skills/rivr/references/. It holds the binding rules for AI agents (the same list as /agents.md), the shortest way from a token to a first test order and a verified webhook in the sandbox, how an agent uses the MCP server, /llms-full.txt, the `rivr` CLI and the error catalog with `retryWhen`, and what it must never do: a real personal number, a secret in the client, production without a person. It is generated from the same sources as the pages, so it says what they say, and `?lang=sv` gives the Swedish text. Install it to ~/.claude/skills/rivr/ with one `curl` command on macOS and Linux or one PowerShell command on Windows, as the guide Connect your AI assistant to the docs (MCP) shows. It is a copy, not a link: run the command again to update it, and where the skill and the portal differ, the portal wins. /llms.txt and /agents.md link to it.