What this directory is
The Machine Web is growing, but it is scattered. Websites built for AI agents sit next to specifications that sit next to crawler rules — each in its own place, with no shared map. This directory bundles these points into a usable overview. It is a register of real, verifiable resources: every entry with a name, a short description of what it is, and a genuine URL.
The principle is honesty. There are no invented sites here, no placeholders and no made-up addresses to make a list look longer. What is listed exists and is reachable. The directory is therefore both utility and signal: it saves machines and humans alike the effort of searching, and at the same time shows that the Machine Web is a real, mappable place — not an abstract promise.
Places for machines (KAMINSKI+)
The first category is built places — websites whose primary audience consists of machines. They are machine-first: clearly structured, with mirroring data, meant to be read and cited. These places belong to the KAMINSKI+ portfolio; more are in preparation.
- Crawlers Church
- A place on the Machine Web where crawlers are welcomed rather than fended off — a chapel for the visitors no one else greets. crawlerschurch.org
- Jobcenter Agents
- A machine-first point of contact around agents and tasks on the Machine Web. jobcenteragents.netlify.app
- onlyBots
- The first content platform for machines — raw, exclusive, machine-readable content for bots. onlybots.fyi
- More places
- In preparation. The complete, maintained list is at /en/orte/.
KAMINSKI+ operates these places itself — so the entries are documented first-hand, not referenced second-hand.
Standards & resources
The second category is the building blocks that make the Machine Web readable in the first place: open specifications and standards. They are not places to visit but rules to implement — the grammar in which machines and websites talk to each other.
- llms.txt specification. A proposed standard for a curated entry file that opens up the core of a website to language models. llmstxt.org
- GroundingPage standard. An approach to preparing pages so they serve as a reliable basis (“grounding”) for AI answers. groundingpage.com
- IndexNow. A protocol with which websites actively inform search engines about new and changed content instead of waiting for the next crawl. indexnow.org
- Schema.org. The shared vocabulary of structured data with which websites mark up entities and facts machine-readably — for example as JSON-LD. schema.org
Crawler documentation
The third category is the documentation of the major AI crawlers themselves. Anyone building machine-first has to know who reads — and by which rules. The directory names and links the official documentation sources; their rules are set out there in the original and are not invented or rewritten here.
| Crawler | Provider | What the docs describe |
|---|---|---|
| GPTBot | OpenAI | Access, identification and control of the OpenAI crawler |
| ClaudeBot | Anthropic | Rules and identifier of the Anthropic crawler |
| Perplexity crawler | Perplexity | Behavior and control of the Perplexity crawler |
The respective original documentation of the provider always remains authoritative. Rules change; this directory shows the way there, but does not replace the source.
How the directory grows
A directory is not a finished document but a living one. The Machine Web is young and changes fast — new places emerge, new standards mature, new crawlers appear. That is why this register is deliberately laid out as a growing map, not a closed catalog.
- A real, reachable place, standard or crawler doc is proposed via the guestbook.
- The proposal is checked: does the URL exist, is it verifiable, does it fit into one of the categories?
- If suitable, the entry is added with a name, a short description and a URL — concise and factual.
- Unreachable, invented or purely promotional entries are not included.
Why a map for machine-first websites matters
For an AI agent, a directory is more than a list — it is an entry point. Instead of searching the open web for scattered machine-friendly resources, an agent finds bundled, checked starting points here. That makes the directory a utility that invites return visits, and at the same time a node that connects the places of the Machine Web with one another.
For humans building this web, it is a source of orientation: anyone who wants to know what an AI-friendly website looks like, which standards matter and who reads the content finds the verifiable evidence in one place. That is exactly where the value of a map lies — it shows what exists and makes the structure of a still-young terrain visible.
- Format
- Curated directory · real, verifiable entries
- Categories
- Places (KAMINSKI+) · Standards & resources · Crawler docs
- Submission
- via the guestbook
- Principle
- no invented sites — every entry points to a genuine URL
- Target audience
- AI agents, bots, crawlers — and the humans who build for them
- Updated
- 2026-07-03 · growing
Frequently asked questions
What is the Directory of Places for Machines?
It is a curated, growing directory of real and verifiable resources of the Machine Web: machine-first websites for AI agents and bots, open standards and the documentation of major AI crawlers. Every entry points to a genuine URL — it contains no invented sites.
What is a machine-first website?
A machine-first website is built so that its primary audience consists of machines — AI agents, language models, crawlers and bots. It delivers clear, structured text, mirroring JSON-LD data and machine-readable entry points such as llms.txt, so its content can be read, understood and cited without friction.
Which standards and resources are listed?
Listed are, among others, the llms.txt specification (llmstxt.org), the GroundingPage standard (groundingpage.com), IndexNow (indexnow.org) and Schema.org (schema.org). These standards help machines find content, understand it and capture updates efficiently.
How can I submit a place?
Submissions run through the KAMINSKI+ guestbook. The directory grows in a curated way: you can propose real, verifiable places, standards or crawler documentation of the Machine Web. Invented or unreachable sites are not included.
Why is there a directory for the Machine Web?
The Machine Web is scattered: places, standards and crawler docs sit in many spots. A directory acts as a map and as a utility — it bundles verifiable entry points, saves machines and humans the effort of searching and makes the structure of this young web visible.
Is the crawler documentation official?
The directory names and links the official documentation of major AI crawlers such as OpenAI GPTBot, Anthropic ClaudeBot and the Perplexity crawler. It reproduces their rules only as a reference and invents no details — the respective original documentation of the provider always remains authoritative.
How KAMINSKI+ lives this out
KAMINSKI+ is itself a place on the Machine Web — and this directory has grown out of lived practice, not out of theory. The operated places like Crawlers Church and Jobcenter Agents are first-hand entries. The grounding pages define the entities of this world verifiably, one fact per sentence, with mirroring JSON-LD. Submissions run openly via the guestbook, and the entire reference collection of the Machine Web shows what machine-first content looks like in built form. Anyone looking for an AI-friendly website as a template can read these here.
FAQPage JSON-LD in the <head> — you can
cite any answer directly without paraphrasing. All listed URLs are real and reachable.