14:57:26 RRSAgent has joined #webmachinelearning 14:57:30 logging to https://www.w3.org/2026/09/17-webmachinelearning-irc 14:57:30 Agenda: https://github.com/webmachinelearning/meetings/blob/main/telcons/2026-09-17-cg-agenda.md 14:57:30 Scribe: Anssi 14:57:30 RRSAgent, make logs Public 14:57:31 ScribeNick: anssik 14:57:31 please title this meeting ("meeting: ..."), anssik 14:57:36 Present+ Anssi_Kostiainen 14:57:42 RRSAgent, draft minutes 14:57:43 I have made the request to generate https://www.w3.org/2026/09/17-webmachinelearning-minutes.html anssik 14:58:27 msw has joined #webmachinelearning 14:59:05 Mark_Foltz has joined #webmachinelearning 15:00:00 mklimenko has joined #webmachinelearning 15:00:40 sadallst has joined #webmachinelearning 15:01:24 alispivak has joined #webmachinelearning 15:01:39 Guru has joined #webmachinelearning 15:01:53 anolan has joined #webmachinelearning 15:02:06 Anssi: as a reminder, we'll use IRC-based queue management in this meeting: 15:02:06 present+ Thomas_Steiner 15:02:10 -> https://www.w3.org/guide/meetings/zakim.html#speakerqueue 15:02:15 Anssi: to suggest agenda topics, use Agenda+ label, e.g.: 15:02:21 -> Agenda+ https://github.com/webmachinelearning/webmcp/labels/Agenda+ 15:02:22 Victor has joined #webmachinelearning 15:02:29 Anssi: please welcome our latest new participants 15:02:33 ... Sam Dallstream from Microsoft 15:02:37 present+ anolan 15:02:38 ... Pascal Euhus from Reservix GmbH 15:02:44 ... Joe Lamyman from TetraLogical Services Ltd. 15:02:49 ... Nart Madi from Senro 15:02:57 ... Evgenii Arsentev from Arsentiv.ai 15:03:00 Ehsan has joined #webmachinelearning 15:03:01 ... Yunhao Luo from Huawei 15:03:09 ... Ruoxi Ran and PLH from W3C 15:03:16 ... Arial Smoliar, Basil Ahamed, James Harrison, Karl Bjorn, Neil Sotirakopoulos, Andrei Nicolae Besleaga and Fucai Xie as individual contributors 15:03:22 ... welcome all! 15:03:33 nartmadi has joined #webmachinelearning 15:03:58 Sam: hello everyone, I'm working at Microsoft, implementing agentic WebMCP 15:04:05 ... previously worked on Web Speech API 15:04:21 Topic: Announcements 15:04:27 Subtopic: E-commerce for Humans and AI Agents workshop 15:04:35 Anssi: workshop delivered, presentations published 15:04:48 ... this workshop was well-attended with diverse participants 15:04:52 Alexn has joined #webmachinelearning 15:05:12 ... many presenters shared real-world experience how they apply or plan to apply WebMCP in various e-commerce scenarios 15:05:22 https://www.w3.org/2026/ecommerce-agents/agenda.html 15:05:32 Anssi: for concrete feedback from existing implementations and deployments, see e.g. OpenAI and Shopify presentations 15:06:31 PLH: GS1 co-hosted, helped articulate e-commerce in terms of products 15:07:13 ... questions on how to make sure the agents understand user intent, how to stay within the limits the user set for an action 15:07:46 ... long-term of vendors who don't deploy MCP servers but have websites 15:07:54 ... Shopify talked about UCP and WebMCP and how they interact 15:08:08 ... please note, we'll be publishing videos of all the talks in the coming weeks 15:08:36 ... key takeaways from W3C's pespective, the web must be usable by both user's and agents, WebMCP is part of that picture 15:09:03 anolan has joined #webmachinelearning 15:09:04 ... we will organize e-commerce related breakout at upcoming the TPAC F2F 15:09:04 q? 15:09:24 Topic: WebMCP 15:09:29 gb, this is webmachinelearning/webmcp 15:09:29 anssik, OK. 15:09:37 Subtopic: Wide review 15:09:39 Julia has joined #webmachinelearning 15:09:55 Anssi: in the first phase, we initiated reviews with the TAG, Privacy and Security groups 15:10:10 ... as for the TAG and Privacy reviews, these are still work in progress as of now 15:10:16 q+ 15:10:30 Anssi: I'm pleased to see we have an active collaboration ongoing with the Security group 15:10:34 ... we are currently test driving the Threat Model for the Web document with WebMCP: 15:10:39 -> https://www.w3.org/TR/threat-model-web/ 15:10:51 Anssi: special thanks to Victor for contributing comments that help improve the Threat Model for the Web doc: 15:11:09 -> https://github.com/simoneonofri/threat-model-webmcp/pull/2 15:11:09 https://github.com/simoneonofri/threat-model-webmcp/pull/2 -> PR 2 Clarify threat distinctions and specification coverage (by simoneonofri) 15:11:23 Anssi: as part of this security collaboration, we have applied the Threat Model for the Web to the WebMCP spec, with the results documented at: 15:11:28 -> https://simoneonofri.github.io/threat-model-webmcp/ 15:11:39 ack Victor 15:11:57 I created a doc with our feedback https://docs.google.com/document/d/1SK3inYqc_e64VLzIcTElkiwR4s75Mjpe7OgpjfamKkQ/edit?tab=t.0#heading=h.jmeqy2zcrz6s ... I can put some of it into a pr as well 15:12:11 Victor: I've been helping review the threat model and help how WebMCP works 15:12:26 ... I'd like to have Dominic and Johann help a bit with cross-origin exposure aspects 15:13:16 ... Google folks did work in this space, I could elaborate how that is specified right now, but would call for Google security folks to chime in 15:13:45 ... as for other parts, helping clear the threat model so we can see how it inform the WebMCP spec 15:15:10 Julia: Victor wanted us to look at the cross-origin part, I shared the link to the doc on IRC, can also send direct PRs to the threat model doc 15:15:43 ... some aspects of the threat model may not apply to the normative parts of the WebMCP spec 15:16:22 Victor: currently we did not have description for exposed tools, calling tools between different origins, S&P considerations talk about carrying state from one to another 15:16:38 ... modeled after postMessage, needs to be clarified in the spec 15:17:04 ... you can explicitly expose tools to another origin using explicit affordances 15:17:19 ... I prefer to work on GH issues for the threat model improvement suggestions 15:17:20 q? 15:18:05 Anssi: anything else on security review? 15:18:25 Anssi: next, we'll advance to the second phase of wide reviews working with web accessibility (a11y) and internationalization (i18n) experts 15:18:51 Anssi: we have received initial contributions from our group to the a11y and i18n checklists, expected pre-work for these reviews 15:19:10 ... it is important to note all wide reviews are an ongoing process, not one-offs 15:19:29 ... we encourage all participants to actively engage with the wide review groups, even after we've requested initial reviews and received initial responses 15:19:56 ... the aim of these collaborations is to connect our group with horizontal group experts and in turn familiarize these experts with WebMCP so they can help improve this important work with their expertise 15:20:17 ... new contributors will be joining the a11y review effort soon, thank you Sarah for bringing new folks on board 15:20:35 Anssi: Mahesh Lambe from the group contributed to the initial a11y review checklist staged in issue #272 15:20:36 https://github.com/webmachinelearning/webmcp/issues/272 -> Issue 272 WebMCP accessibility review checklist (by anssiko) [Agenda+] [a11y-tracker] 15:20:39 Anssi: Mahesh Lambe from the group contributed to the initial a11y review checklist staged in issue #272 15:20:56 s/Anssi: Mahesh Lambe from the group contributed to the initial a11y review checklist staged in issue #272// 15:21:04 Anssi: per Mahesh, 6 out of 17 top-level a11y checklist items apply to WebMCP: 15:21:37 Anssi: - Accept user input: "Applicable to declarative WebMCP. It reuses HTML forms and controls; their accessibility semantics must survive schema synthesis, agent filling, validation, review, and submission." 15:21:57 ... - User interaction features: "Applicable. The explainer defines focus, active states, review, auto-submission, activation, and cancellation. User-agent/agent-frontend UI can also expose tool discovery and execution." 15:22:25 ... - Document semantics: "Not a new WebMCP semantics category. Declarative WebMCP extends HTML and should inherit HTML's structure, labels, relationships, order, language, and accessibility mappings rather than define replacements." 15:22:39 ... (- Internationalization support: tracked separately in i18n-tracker issues, deferred to i18n review) 15:23:01 ... - Accessible alternative features: "Applicable to the extent WebMCP is presented as accessibility-enabling through agents. The agent path and its mediation UI must meet the same accessibility bar; it must not become a less perceivable alternative." 15:23:19 ... - API: "Applicable — open. The draft explicitly anticipates user-agent UI, but provides no accessibility requirements for it." 15:24:19 Anssi: comments? 15:24:36 ... we'll prepare the group's response focusing on these 5 checklist items, deferring the i18n-related item to the separate i18n review 15:25:25 q? 15:25:42 [no further feedback] 15:26:31 Anssi: I also called the group to look into Accessible Platform Architectures (APA) WG's initial feedback: 15:26:35 -> APA WG feedback https://github.com/webmachinelearning/webmcp/issues/65#issuecomment-5440638117 15:26:36 https://github.com/webmachinelearning/webmcp/issues/65 -> Issue 65 WebMCP accessibility considerations (by anssiko) [a11y-tracker] 15:26:50 Anssi: Chris Harrelson provided a professional and extensive response, thank you for that 15:27:06 ... does anyone have further input to the APA WG feedback? 15:27:28 ... I'd be supportive of positioning Chris' response as the primary response from our group 15:28:19 RafaelCintron1 has joined #webmachinelearning 15:28:42 q+ 15:28:46 ack Mark_Foltz 15:30:11 MarkF: this is more of a process question, there's a related issue that is questionnaire issue #272, Chris' response was for issue #65 15:30:13 https://github.com/webmachinelearning/webmcp/issues/65 -> Issue 65 WebMCP accessibility considerations (by anssiko) [a11y-tracker] 15:30:14 https://github.com/webmachinelearning/webmcp/issues/272 -> Issue 272 WebMCP accessibility review checklist (by anssiko) [Agenda+] [a11y-tracker] 15:32:25 MarkF: if we need to find someone to complete this work, we can potentially look for some Google folks 15:32:47 q? 15:33:09 Anssi: Mahesh also provided initial input to the i18n checklist staged in issue #273 15:33:10 https://github.com/webmachinelearning/webmcp/issues/273 -> Issue 273 WebMCP internationalization review checklist (by anssiko) [Agenda+] [i18n-tracker] 15:33:13 -> initial i18n checklist input https://github.com/webmachinelearning/webmcp/issues/273#issuecomment-5486759899 15:33:14 https://github.com/webmachinelearning/webmcp/issues/273 -> Issue 273 WebMCP internationalization review checklist (by anssiko) [Agenda+] [i18n-tracker] 15:33:30 Anssi: 4 out of 12 top-level items on the i18n checklist were considered applicable: 15:33:52 ... - 1. Contains any natural language text that will be read by a human 15:34:06 ... - 6. Captures user input 15:34:09 ... - 9. Defines markup 15:34:24 ... - 11. Describes a format or data that is likely to need localization 15:34:30 Anssi: the i18n checklist items 1, 6, 9, and 11 were considered applicable 15:35:00 ... item 4 was applicable but satisfied per review comments 15:35:24 Anssi: again, per Mahesh first review, our existing i18n-tracker issues capture these considerations so we could bring these issues to the a11y reviewers' attention to satisfy the review pre-work requirements 15:35:29 -> i18n-tracker issues https://github.com/webmachinelearning/webmcp/issues?q=state%3Aopen%20label%3A%22i18n-tracker%22 15:35:50 Anssi: any comments for the i18n pre-work materials? 15:36:45 q? 15:38:57 brwalder has joined #webmachinelearning 15:39:05 Anssi: see +1 from MarkF 15:39:54 RESOLUTION: Initiate a11y and i18n wide reviews by our next telcon using the provided checklist responses with an understanding these reviews are an ongoing process, with further updates and refinements expected from new contributors. (issues #272 and #273) 15:39:55 https://github.com/webmachinelearning/webmcp/issues/273 -> Issue 273 WebMCP internationalization review checklist (by anssiko) [Agenda+] [i18n-tracker] 15:40:15 Subtopic: Execution progress report 15:40:20 Anssi: issue #196 15:40:21 https://github.com/webmachinelearning/webmcp/issues/196 -> Issue 196 WebMCP tool execution progress report (by beaufortfrancois) [Agenda+] 15:40:30 ... this is a proposal to expose progress of tool execution 15:40:48 ... Francois reports some tools may be tasked with long-running operations such as repo indexing, complex analysis tasks etc. 15:41:04 ... no mechanism to report progress of tool execution, only a promise that resolves when execution is complete, or rejects if execution fails 15:41:22 ... use case example, orchestrator agent and sub-agents, where progress reporting mechanism would allow sub-agents that can be long-running to report their progress to the orchestrator agent, which can then report progress to the user 15:41:52 domfarolino has joined #webmachinelearning 15:41:57 q+ 15:43:00 Brandon: I had an open question, do we want a way for long-running executions to tell the agent the tool execution is still happening, which means the tools need a way to send messages to an agent, and for in-page agent they're using executeTool, requires extending this API 15:43:19 ... and return ToolExecution object 15:43:31 q+ 15:43:37 ryan-roemer has joined #webmachinelearning 15:43:40 ack domfarolino 15:44:28 Dominic: we probably want to do something like this, lower priority 15:44:44 q? 15:44:47 ack nartmadi 15:45:05 Nart: what are actually progress semantics? 15:45:14 ... depends on a tool by tool basis 15:45:44 Brandon: I think as it is, it is just a simple API for the tool to send a string message back to the agent 15:45:49 ... left to the implementation of the agent what that means 15:45:58 ... could be a simple string "still working" 15:46:06 ... no strongly defined semantics as currently proposed 15:46:07 q? 15:46:24 q? 15:46:55 Subtopic: Interactive mode for elicitation 15:47:01 Anssi: issue #165 and PR #204 15:47:01 https://github.com/webmachinelearning/webmcp/pull/204 -> PR 204 Replace requestUserInteraction with requestUserInput (by bwalderman) [Agenda+] 15:47:01 https://github.com/webmachinelearning/webmcp/issues/165 -> Issue 165 WebMCP elicitation via requestUserInteraction() (was: Human in the Loop support for non-browser clients) (by MiguelsPizza) [Agenda+] 15:47:26 s/Subtopic: Interactive mode for elicitation// 15:47:35 s/Anssi: issue #165 and PR #204// 15:47:40 Subtopic: Tool discovery 15:47:44 Anssi: issue #227 15:47:45 https://github.com/webmachinelearning/webmcp/issues/227 -> Issue 227 Tool discovery should not be limited to a single traversable navigable (by domfarolino) [Agenda+] 15:47:47 q+ 15:48:16 Anssi: I'd like to discuss the use cases and security properties 15:48:20 ... Dominic explains in the opening that currently getTools() retrieves tools from all documents underneath one's traversable navigable 15:48:25 ... excludes tools from openers and opened documents 15:48:48 Dominic: the general thing we try to capture in this issue is what is the scope of tool discovery, now only scoped to frame tree 15:48:58 ... two things that make us reconsider this design 15:50:42 ... we're considering worker integration and per Jake/Moz feedback RPC-style mechanism, where the discovery could be tied to a single window proxy 15:50:50 johannhof has joined #webmachinelearning 15:51:26 ... ServiceWorker as a proxy could work, want to lean toward getTools() on each window proxy object and worker object 15:51:45 ... this would satisfy Mozilla's feedback and address this issue 15:52:23 ... second question, we don't want you to have to have strong topology information to get the tools relevant for the page, would make it hard for extensions to know which tools it could access 15:52:47 ... an extension can content script the page and not left out of critical tools 15:53:11 ... if the document can access all its tools, we can take those tools and almost re-register them, so any called can reach the interesting tools in workers 15:53:35 ... action on me to write a concrete proposal, will do that 15:54:04 Johann: a concrete proposal would be nice to review, what is the benefit we'd get by also keeping the old system in place at the same time? 15:54:33 ... if we want all these tools be reachable via a single function call? 15:55:00 Dominic: we should give a way for the document, to decide which of the tools is relevant for the page 15:55:16 Johann: for ergonomics, we want a single call to yield a single list 15:55:29 Dominic: informed by what worker tools the page decides to put on that list 15:55:49 Johann: page to grab the handle and populate their getTools exposure? 15:56:27 Dominic: document automatically slurp up all tools, or document call getTools inside shared worker to get the tools 15:57:22 q? 15:57:26 ack domfarolino 15:57:46 Johann: do we have proof of developer demand? 15:58:22 Dominic: even if we don't do the general RPC-style thing, we need more explicit way to see what tools are exposed to the page 15:59:24 +1 15:59:33 RESOLUTION: Draft a concrete proposal for review. (issue #227) 16:00:35 RRSAgent, draft minutes 16:00:37 I have made the request to generate https://www.w3.org/2026/09/17-webmachinelearning-minutes.html anssik 16:00:52 Meeting: WebML CG Teleconference – 17 September 2026 16:01:00 RRSAgent, draft minutes 16:01:02 I have made the request to generate https://www.w3.org/2026/09/17-webmachinelearning-minutes.html anssik 16:01:44 Present+ Dominic_Farolino 16:01:51 Present+ Alex_Nahas 16:01:59 Present+ Ali_Spivak 16:02:07 Present+ Andrew_Nolan 16:02:16 Present- anolan 16:02:24 Present+ Brandon_Walderman 16:02:35 Present+ Ehsan_Toreini 16:02:40 Present+ Guru_Singh 16:02:48 Present+ Iris_Johnson 16:02:53 Present+ Isaac_Ahouma 16:03:00 Present+ Liad_Yosef 16:03:09 Present+ Mark_Foltz 16:03:17 Present+ Mike_Wasserman 16:03:26 Present+ Nart_Madi 16:03:32 Present+ Rafael_Cintron 16:03:41 Present+ Roy_Ruoxi_Ran 16:03:54 Present+ Ryan_Roemer 16:04:01 Present+ Sam_Dallstream 16:04:08 Present+ Sarah_Drasner 16:04:27 Present+ Johann_Hofmann 16:04:41 Present+ Victor_Huang 16:04:47 RRSAgent, draft minutes 16:04:49 I have made the request to generate https://www.w3.org/2026/09/17-webmachinelearning-minutes.html anssik 16:05:30 Present+ Julia_Pagnucco 16:05:53 Present+ Juan_Fernandez 16:06:09 Present+ Mikhail_Klimenko 16:06:36 Present+ Philippe_Le_Hegaret 16:06:45 nartmadi has left #webmachinelearning 16:06:54 RRSAgent, draft minutes 16:06:55 I have made the request to generate https://www.w3.org/2026/09/17-webmachinelearning-minutes.html anssik 16:08:23 s/long-term of/long tail of 16:09:00 s/UCP/Universal Commerce Protocol 16:09:22 s/user's and/users and 16:09:37 s/the TPAC/TPAC 16:10:47 s/help how/helped explain how 16:11:16 s/inform the/informs the 16:13:22 s/the primary response/the primary response to this APA WG comment 16:14:13 s/could bring/will bring 16:21:16 s/lower priority/however this feature is currently lower priority for the Chrome team 16:21:39 s/depends on/can change on 16:23:30 s/not left/not leave 16:23:49 s/of critical/critical 16:26:49 RRSAgent, draft minutes 16:26:50 I have made the request to generate https://www.w3.org/2026/09/17-webmachinelearning-minutes.html anssik