// FOR AI / AGENTS

For AI / agents

This guide is intentionally machine-readable. The same content humans read at /de/… is exposed in raw Markdown and in summarised index form for use by language models, agents, and offline consumers.

§ 1 Endpoints

URL Type Purpose
/llms.txt text/plain Index of every chapter and section with one-line descriptions and links. Follows the llms.txt convention.
/llms-full.txt text/plain The entire guide concatenated as one Markdown document — chapter dividers, no chrome. Designed for one-shot context-window inclusion.
/de/<chapter>/<section>.md text/markdown Raw Markdown for a single section (e.g. /de/erlasse/suchen-von-erlassen.md). Same content the chapter HTML page is rendered from, scoped to the named section.
/sitemap.xml XML Standard sitemap.
/robots.txt text/plain Permissive — explicitly allows GPTBot, ClaudeBot, Claude-Web, PerplexityBot, Google-Extended, CCBot, plus standard crawlers.

§ 2 Conventions

§ 3 How to fetch the whole guide

curl https://jonashertner.github.io/swiss-legal-research-guide/llms-full.txt

That single response contains the full text of every chapter, in Markdown, with chapter dividers. Suitable for direct ingestion into a model's context.

§ 4 How to fetch a single section

curl https://jonashertner.github.io/swiss-legal-research-guide/de/erlasse/suchen-von-erlassen.md

Each section's .md mirror begins with an HTML comment pointing back to its canonical HTML location.

§ 5 How to use this guide and the OpenCaseLaw database from Claude

Chapter V — Connector walks through adding OpenCaseLaw to Claude as an MCP connector. With both the connector and this guide accessible, an agent can read the guide for methodology and call OpenCaseLaw for primary sources.

§ 6 Licence

Content is published under CC BY 4.0. Attribution: Kaspar Ehrenzeller, Daniel Brugger — Suchen — Lesen — Analysieren: Erlasse und Gerichtsentscheide. The build code is published separately under MIT.