Re: WebMCP

Here arer my thoughts on the concept:

Web MCPIt may become important because AI agents could assist disabled
users by performing tasks on their behalf. In that sense, WebMCP could
become part of an “agent accessibility” layer.  Still, WebMCP may duplicate
the accessibility tree and create a parallel description of page actions
that can drift out of sync with the real DOM/accessibility layer.

How would this work if more closely associated with the existing
accessibility tree framework and function as a plain language extension?

*Dr.  Keith Newton *(Call me "Dr. Keith")
Executive Leader and Strategic a11y in Digital Accessibility Experience
614-654-4116 <%28614%29%20654-4116> |
https://www.linkedin.com/in/dr-keith-newton  | newtonsgroove2@gmail.com


On Mon, Jun 29, 2026 at 6:01 PM <m.atkinson@samsung.com> wrote:

> Dear colleagues,
>
> The W3C Team (staff) have asked for our input to the discussion about
> whether WebMCP should be adopted as official W3C work. We have discussed
> different aspects of this a couple of times recently in the main APA WG
> call, and I'm in the process of tying those threads together, and hope to
> have one possible proposed comment for your consideration by this week's
> main APA WG meeting.
>
> However, regardless as to whether you attend the main APA WG meetings or
> not, it would be great if you could read some of the following, and share
> your thoughts. I will try to reflect them in any proposed comment that I
> submit for APA WG's consideration. We will be discussing these again on
> this
> week's main APA WG call.
>
> Perhaps the key thread is the one regarding the relationship between WebMCP
> and the accessibility tree - there's a fundamental question as to whether
> the accTree is the right place to add semantics, and there are concerns on
> either side as to the results of doing, or not doing, so.
>
> Note for people who don't regularly attend APA WG review calls: the links
> to
> 'tracking threads' are links to issues in APA's space, where APA members
> are
> encouraged to natter and co-ordinate, before proposing comments. To get to
> the source issues, follow the link from the first comment of the tracking
> issue. The benefit of the layer of indirection is we can talk about an
> issue, and develop APA consensus around it, before commenting as APA WG on
> the public thread. You're still welcome to comment on the public threads as
> individuals.
>
> * WebMCP explainer:
> <https://github.com/webmachinelearning/webmcp/blob/main/README.md>
> * Tracking issue for WebMCP accessibility considerations:
> <https://github.com/w3c/a11y-tracking/issues/318>
> * Tracking issue for 'Redundancy with accessibility tree' thread:
> <https://github.com/w3c/a11y-tracking/issues/317>
> * Standards positions of the three main engines:
>   - Chromium/Chrome: Supportive
>   - Gecko/Mozilla Firefox: No official position yet; thread:
> <https://github.com/mozilla/standards-positions/issues/1412>
>   - WebKit/Safari: Opposed; position thread:
> <https://github.com/WebKit/standards-positions/issues/670> and the
> specific
> comment where they state their objections is:
> <
> https://github.com/WebKit/standards-positions/issues/670#issuecomment-46084
> 32694>
>
> The explainer starts with a good intro as to what WebMCP is.
>
> Thanks for your time, and any input you may have! Feel free to share via
> this email, or via the a11y-tracking issues linked above (on the threads in
> the a11y-tracking repo, not the source repos).
>
> best regards,
>
>
> Matthew
>
>
>