The Linux Foundation starts ANS, a name system for AI agents
ANS shows who an AI agent is, but it does not tell the tasks that the agent can do.
On 10 November 2026, the Linux Foundation made Agent Name Service (ANS) a project of the Foundation, with an open specification and open source tools. ANS lets a system find which AI agent sends a message, before it uses its policy for agents. Since 23 June 2026, 11 organizations work on ANS or use it.
How it works
An ANS name holds a version and a domain that the owner of the agent controls. A registry checks the domain, and the Domain Name System (DNS) of the internet holds the discovery data. An open ledger records each entry with a digital signature.
The proof
The Foundation gives open source software in Rust, Java and the Go language, with no test numbers or use numbers. The quickstart uses test keys and a DNS check that does not look at the open DNS. A good result in it does not show that a live system works.
What is not known
The draft for agent authentication is version 0.2.0 and does not include permissions, delegation or trust scores. A receipt shows the history of the ledger, not the status at this time. It can show a good result after the registration of an agent ends.
Insight: Many groups now try to own the open standard for agents
In October 2026, Anthropic, OpenAI and Alibaba gave agent tools to one neutral foundation. On 10 November 2026, NVIDIA gave OpenShell to the CNCF, and ANS became a Linux Foundation project. This reads as a move to get the first standard that vendors accept, and that standard can set the rules. ANS includes identity only. You must make your permission policy.
Sources
Posted
