Assistants that reach into systems you have to answer for.
Anthropic builds Claude and authored the Model Context Protocol, which it donated to the Linux Foundation. Our work is scoping what assistants reach, the connection layer, and the identity work underneath agent access.
We build on Anthropic's platform where agents connect to the governed estate. The connection is standardized now; what it is allowed to see is still a permission question.
What it is
Anthropic builds Claude and authored the Model Context Protocol, which it donated to the Linux Foundation. That protocol is now how assistants connect to systems of record, including the Atlassian estate we govern.
We build on that rather than reselling it. Our work is on the side of the connection the customer is accountable for: which identity an assistant acts as, what that identity may reach, and what evidence exists afterwards.
Where it fits
Anthropic sits at L4, agents and automation, and depends entirely on L1. An assistant sees exactly what the identity behind the connection permits. Standardizing the protocol removed the integration problem and left the governance one, which is the layer we work in.
What we do with it
Scope the connecting identity
An assistant connects as an account. Designing that account, rather than reusing an administrator credential, is the control.
Usage policy and guardrails
What may be connected, by whom, and what record is kept when it acts.
Build what the standard does not cover
Where the estate needs behavior the standard connection does not provide, we build it inside the permission model.
Proof
Monthly SDK downloads of the Model Context Protocol, donated to the Linux Foundation in December 2025.
Anthropic and the Linux Foundation, December 2025.
AI Innovator Finalist, 2026
Atlassian Partner Awards, 2026