AI Docs Standard (documentation.ai.md)
Open standard for a single Markdown file that lets AI agents install, configure, and operate your software without a human reading the docs first.
🔗 Visit AI Docs Standard (documentation.ai.md)
Description
Regular documentation is written for a human who can infer context, skip irrelevant sections, and ask a colleague when something's unclear. An AI agent trying to install and configure your software from a README doesn't have any of that — it needs exact commands, exact ports, exact response shapes, with nothing left implicit. documentation.ai.md is a proposed standard for exactly that: one Markdown file per product that's written for a machine reader, not a person.
The standard defines a fixed structure — identity, install steps, configuration, an interface quickstart, an admin surface, architecture facts, and links — all formatted around copy-pasteable commands, explicit environment variables, exact port numbers, and exact response shapes, so an agent can parse the file and act on it directly rather than guessing. It's meant to complement the llms.txt standard (which describes a site for LLM consumption) rather than replace it — documentation.ai.md focuses specifically on operational instructions for installing and running software. It's a single English-language file per product/version, licensed CC BY 4.0, with reference implementations already shipped in the maintainer's own products.
💬 Our review
The short version: this is a genuinely useful idea at an early stage — worth adopting if you're building software you want AI agents to be able to install and operate on their own, but it's a proposal from a single maintainer, not yet an industry-wide standard, so don't expect broad tooling support around it today.
The closest comparison is llms.txt, which is already seeing wider adoption for describing a website's content to LLMs; documentation.ai.md is narrower and more operational, focused on install/configure/run instructions rather than general content description, so the two are complementary rather than competing. Since it's free (CC BY 4.0, just a file format spec) there's no cost tradeoff — the real question is adoption risk: writing a documentation.ai.md file today is cheap, but whether agents widely start looking for this specific filename is unproven. Worth it as a low-cost bet if you maintain a self-hosted tool and want to be ahead of the curve; not worth the effort yet if your users are entirely human. <!-- ai-generated -->
💰 Pricing
📊 Global score
🤖 AI-enriched data
Standard ouvert (CC BY 4.0), gratuit, aucun coût
Pros
structure fixe pensée pour un parsing machine, pas humain
complémentaire à llms.txt plutôt que concurrent
gratuit, un seul fichier Markdown à ajouter au projet
déjà utilisé en interne dans les produits du mainteneur (Kivi, Vesta)
Cons
standard porté par un seul mainteneur, pas encore une norme largement adoptée
aucun outillage/validateur tiers connu à ce stade
valeur dépend de l'adoption future par les agents IA eux-mêmes