14:53:32 RRSAgent has joined #webmachinelearning 14:53:36 logging to https://www.w3.org/2026/10/01-webmachinelearning-irc 14:53:36 RRSAgent, make logs Public 14:53:37 please title this meeting ("meeting: ..."), anssik 14:53:49 Meeting: WebML CG Teleconference – 1 October 2026 14:53:49 Chair: Anssi 14:53:49 Meeting: WebML CG Teleconference – 1 October 2026 14:53:53 Agenda: https://github.com/webmachinelearning/meetings/blob/main/telcons/2026-10-01-cg-agenda.md 14:53:58 Scribe: Anssi 14:54:04 ScribeNick: anssik 14:54:09 Present+ Anssi_Kostiainen 14:54:17 RRSAgent, draft minutes 14:54:18 I have made the request to generate https://www.w3.org/2026/10/01-webmachinelearning-minutes.html anssik 14:57:36 Mark_Foltz has joined #webmachinelearning 14:58:37 Present+ Mark_Foltz 14:59:06 Present+ Jeff_Whelpley 14:59:30 jeffwhelpley has joined #webmachinelearning 15:00:07 Present+ Dominic_Farolino 15:00:52 Present+ Philippe_Le_Hegaret 15:01:05 ryan-roemer has joined #webmachinelearning 15:01:32 RRSAgent, draft minutes 15:01:33 I have made the request to generate https://www.w3.org/2026/10/01-webmachinelearning-minutes.html anssik 15:02:21 ali has joined #webmachinelearning 15:02:37 domfarolino has joined #webmachinelearning 15:02:47 Ehsan has joined #webmachinelearning 15:02:52 Anssi: as a reminder, we'll use IRC-based queue management in this meeting: 15:02:58 -> https://www.w3.org/guide/meetings/zakim.html#speakerqueue 15:03:03 tidoust has joined #webmachinelearning 15:03:09 Anssi: to suggest agenda topics, use Agenda+ label, e.g.: 15:03:19 plh has joined #webmachinelearning 15:03:20 -> Agenda+ https://github.com/webmachinelearning/webmcp/labels/Agenda+ 15:03:33 Anssi: please welcome our latest new participants 15:03:33 present+ 15:03:42 ... Nathan Schloss from Meta 15:04:04 ... Idan Levin from nekuda 15:04:23 BenGreenstein has joined #webmachinelearning 15:04:29 present+ Thomas_Steiner 15:04:39 ... Sohail Hooda from LogMeIn Inc. 15:04:55 ... Ian Holmes, Eyup Ucmaz, Melvin Carvalho as individual contributors 15:04:59 ... welcome all! 15:05:02 anolan has joined #webmachinelearning 15:05:20 Topic: Announcements 15:05:28 Subtopic: WebMCP polyfill 15:05:35 Anssi: WebMCP polyfill project has been launched 15:05:56 ... congratulations to the team with special thanks to Alex Nahas and Francois Beaufort for the initial contributions to establishing the project 15:06:00 -> https://github.com/webmachinelearning/webmcp-polyfill 15:06:30 Anssi: this WebMCP polyfill project directly supports W3C's spec work and is developed in an open and transparent manner 15:06:42 ... and thus qualifies as an official W3C Open Source Program pilot project 15:06:53 Anssi: I have invited Francois Daoust who co-leads W3C's Open Source Program to tell more about the pilot and next steps 15:08:03 Francois: three high-level things, the scope of OSS Program is to recognize W3C is always grounded on open source, we like to embrace that and get more things out that work hand in hand with W3C standards 15:08:14 ... WebMCP polyfill is 100% aligned with that 15:08:55 ... I saw discussion around naming of an npm package, so @w3c/webmcp-polyfill would be available to the project in this pilot 15:08:55 https://github.com/w3c -> @w3c 15:09:37 Francois: during this pilot, we are working closely with the projects that are onboarding 15:10:25 Anssi: thank you Francois for the guidance 15:10:31 Anssi: I will work with the polyfill team on practicalities to advance the WebMCP polyfill project into the W3C OSS Program pilot project 15:11:18 Subtopic: WebMCP momentum and next steps 15:11:41 Anssi: we're all heads down working so it is easy to overlook the remarkable growth and achievements of everyone working on WebMCP in this group 15:11:51 Anssi: first, growth in the WebMCP ecosystem is remarkable 15:12:07 ... this Community Group that initiated the WebMCP effort a year ago has grown +55% YoY, we're currently 314 participants across 84 organizations 15:12:11 -> https://www.w3.org/groups/cg/webmachinelearning/participants/ 15:12:14 q+ to briefly mention the WebMCP extension 15:12:26 Anssi: the WebMCP repo has grown from 0 to 4.4k GH stars YoY, remarkable for a web spec: 15:12:31 -> https://github.com/webmachinelearning/webmcp 15:12:48 Anssi: the recent WebMCP Challenge attracted over 5k submissions from developers: 15:12:55 -> https://openai.com/webmcp-challenge/ 15:13:11 Anssi: the broader web community continues to impress us with their engagement, contributions, and innovative use cases for WebMCP, some of which are captured in the Awesome WebMCP repo: 15:13:14 sushraja-msft has joined #webmachinelearning 15:13:19 -> https://github.com/webmachinelearning/awesome-webmcp 15:13:48 Anssi: WebMCP implementation experience is growing steadily and most importantly diversifying to different platforms and environments: 15:13:50 -> https://github.com/webmachinelearning/webmcp/blob/main/implementation-status.md 15:14:14 Anssi: during this rapid growth phase, this Community Group has proactively engaged with various stakeholders to ensure this growth of the WebMCP ecosystem happens in a responsible manner, focused on privacy, security, the needs of humans and agents 15:14:41 ... the group is actively working with W3C's horizontal groups; the TAG, Privacy, Security, Accessibility, and Internationalization, helping these groups understand WebMCP and in turn improve WebMCP in these critical areas 15:15:00 ... I want to thank all of you for your dedication and contributions to the WebMCP effort 15:15:04 Anssi: I've invited W3C's Philippe Le Hégaret who's responsible for W3C's tech strategy to outline the next steps on the agentic web journey 15:15:28 q+ 15:16:01 Philippe: as we all know, humans are relying on AI agents to do their work, on the web, and we hope it matches the user intent, we want the web to be accessible to agents, WebMCP allows actuation by AI agents 15:16:02 https://w3c.github.io/charter-drafts/2026/agenticweb-wg.html 15:16:22 ... we announced yesterday we're working on Agentic Web WG to deal with that problem space 15:16:34 ... intersection of agents and web interactions 15:16:54 ... this is an early draft, I expect conversations at TPAC, maybe in this CG 15:17:18 ... TPAC breakouts have many proposals that are in the space of agentic web 15:17:53 ... the idea would be to WebML WG would focus on ML in the browser, and this proposed WG would focus on agent interactions on the web 15:18:17 ... potentially also best practices docs for developers to help them design their web sites for better consumption by agents 15:18:21 q? 15:18:27 ack domfarolino 15:19:07 Dominic: thanks for getting this off the ground, we've been interested in the venue, to have security and harness issues focus, this sounds a nice way to also work on those 15:19:25 ... how do I join, the charter draft join button is not working 15:19:39 Philippe: it is just a draft, it will take a few months to set this up 15:20:14 ... some issues you have, need some incubation, so we expect incubution, using WebML CG for incubating those ideas prior 15:20:37 Dominic: one of the deliverables is WebMCP, I guess the plan is to adopt it 15:20:47 Philippe: that is the current plan 15:20:54 q? 15:21:19 https://github.com/w3c/strategy/issues/576 15:21:19 Johann: can you paste the link how to follow along the process? 15:21:19 https://github.com/w3c/strategy/issues/576 -> Issue 576 [wg/agenticweb] Agentic Web Group Charter (by plehegar) [AI] [In Charter Refinement] [charter] 15:21:29 johannhof has joined #webmachinelearning 15:21:40 https://github.com/w3c/strategy/issues/576 15:21:43 Philippe: that's the link to the strategy issue 15:21:44 q? 15:21:54 Alex8 has joined #webmachinelearning 15:22:01 iris has joined #webmachinelearning 15:22:09 Philippe: I can organize a breakout and discuss during the WebML CG TPAC F2F this proposal 15:22:12 IsaacAhouma has joined #webmachinelearning 15:22:30 Philippe: the plan is we've be happy with the charter by TPAC end 15:22:34 q? 15:22:37 ack tomayac 15:22:38 tomayac, you wanted to briefly mention the WebMCP extension 15:22:44 (sorry for jumping you tom) 15:22:58 https://github.com/googlechromelabs/webmcp-extension 15:23:15 Tom: on WebMCP, brief announcement, our team, we introduced WebMCP extension 15:23:28 ... working in the Prompt API and WebMCP intersection 15:23:29 q? 15:24:17 q? 15:24:26 In the extension, you can choose the Prompt API to back the WebMCP tool calling. 15:24:43 Philippe: keep up the great work on WebMCP! 15:25:30 Dominic: one thing that makes the WG successful is when we define normative requirements for browser use agent harness, have frontier labs in the room 15:26:04 Philippe: I have a call with OpenAI next week, they presented at W3C workshop last month 15:26:39 Topic: WebMCP 15:26:43 gb, this is webmachinelearning/webmcp 15:26:43 anssik, OK. 15:26:51 Subtopic: Wide review 15:27:05 Anssi: per our resolution, we have initiated web accessibility (a11y) and internationalization (i18n) reviews: 15:27:11 -> WebMCP a11y review request https://github.com/w3c/a11y-request/issues/189 15:27:12 https://github.com/w3c/a11y-request/issues/189 -> Issue 189 WebMCP 2026-10-01 (by anssiko) [SR] [WD] [REVIEW REQUESTED] [pending] [s:webmcp] 15:27:19 -> WebMCP i18n review request https://github.com/w3c/i18n-request/issues/338 15:27:20 https://github.com/w3c/i18n-request/issues/338 -> Issue 338 WebMCP 2026-10-01 > 2026-10-27 (by anssiko) [REVIEW REQUESTED] [SR] [pending] 15:27:39 Anssi: as part of these review requests, we have provided as references a11y and i18n checklists the group prepared: 15:27:43 -> a11y checklist https://github.com/webmachinelearning/webmcp/issues/272 15:27:44 https://github.com/webmachinelearning/webmcp/issues/272 -> Issue 272 WebMCP accessibility review checklist (by anssiko) [Agenda+] [a11y-tracker] 15:27:48 -> i18n checklist https://github.com/webmachinelearning/webmcp/issues/273 15:27:48 https://github.com/webmachinelearning/webmcp/issues/273 -> Issue 273 WebMCP internationalization review checklist (by anssiko) [Agenda+] [i18n-tracker] 15:28:08 Anssi: as discussed, both of these reviews and related issues remain open for additional community feedback 15:28:19 Victor has joined #webmachinelearning 15:28:21 ... these wide reviews are an ongoing process, not one-offs 15:28:38 ... any comments, questions about a11y and i18n reviews so far? 15:28:39 q? 15:29:14 Anssi: in June, we reached out to the Privacy WG for privacy review and feedback on WebMCP's privacy implications: 15:29:19 -> https://github.com/w3cping/privacy-request/issues/213 15:29:20 https://github.com/w3cping/privacy-request/issues/213 -> CLOSED Issue 213 WebMCP 2026-06-25 > 2026-08-01 (by domfarolino) [WD] [REVIEW REQUESTED] 15:29:23 Anssi: the Privacy WG group first discussed WebMCP on its 16 July call: 15:29:30 -> https://github.com/w3c/privacywg/blob/main/minutes/2026/privacywg-20260716.md 15:29:42 Anssi: last week, two privacy-related issues were opened for WebMCP for discussion: 15:29:56 -> https://github.com/webmachinelearning/webmcp/issues/313 15:29:56 https://github.com/webmachinelearning/webmcp/issues/313 -> Issue 313 Mitigations for profound privacy and security issues seem to be missing? (by npdoty) [Agenda+] [privacy-needs-resolution] 15:30:00 -> https://github.com/webmachinelearning/webmcp/issues/313 15:30:10 Anssi: today, I'd like to discuss these two issues to converge on consensus on the group's response 15:30:30 q+ 15:30:33 Anssi: first, in issue #313, two pieces of feedback: 15:30:33 https://github.com/webmachinelearning/webmcp/issues/313 -> Issue 313 Mitigations for profound privacy and security issues seem to be missing? (by npdoty) [Agenda+] [privacy-needs-resolution] 15:30:48 Julia has joined #webmachinelearning 15:30:52 Anssi: i) an observation that an agent with broad authority can create risks 15:31:05 ... ii) suggestion to consider threats where the site author is the attacker 15:31:28 ... for the first, the initial response from Dominic suggests this concern is relevant regardless of WebMCP and would be best addressed on a different layer 15:31:42 q+ 15:31:45 ... for the second, I wonder if some non-normative "Guidelines for Agent Harness Implementers", possibly a separate document, or deliverable, would be helpful 15:31:48 q+ 15:31:52 ... Google folks have published materials that could feed into these guidelines to turn this into community effort: 15:31:58 https://developer.chrome.com/docs/agents/security 15:31:58 -> Agent dev guidance https://developer.chrome.com/docs/agents/security 15:32:03 -> Site dev guidance https://developer.chrome.com/docs/ai/webmcp/secure-tools 15:32:14 Anssi: thoughts on the privacy feedback, specifically these two issues #313 and #314? 15:32:15 https://github.com/webmachinelearning/webmcp/issues/314 -> CLOSED Issue 314 visible UI and forms might have fewer open-ended prompt injection attacks (by npdoty) [Agenda+] [security-tracker] [privacy-needs-resolution] 15:32:15 q? 15:32:22 ack domfarolino 15:32:29 Dominic: well covered, Anssi 15:32:51 ... we've been using a new label, "beyond scope agent identity" as a collection of issues that should be solved in Agentic Web WG scope 15:33:13 ... I think it'd be great to point to the resources Julia shared and work on them 15:33:13 q? 15:33:14 q+ 15:33:16 ack Victor 15:33:38 Victor: it seems the concerns from Privacy WG are similar to discussion we had with Security WG 15:34:00 ... perhaps we could collapse the both groups and work on both privacy and security considerations together 15:34:11 sushraja-msft has joined #webmachinelearning 15:34:38 ... prompt injections is something we've considered, also need to discuss this with both security and privacy groups present 15:34:48 q? 15:34:51 ack Julia 15:35:38 +1 15:35:42 Julia: I agree with Victor and Dom, agent security is a bit out of scope as of now, we say that is pending for new WG where to discuss those aspects 15:35:42 q? 15:35:46 ack Mark_Foltz 15:36:16 MarkF: one topic which is someone in scope of this group is issue #173 15:36:17 https://github.com/webmachinelearning/webmcp/issues/173 -> CLOSED Issue 173 Make tools Document-scoped instead of Window-scoped (by domfarolino) 15:36:38 ... guardrails for harnesses, internal discussions around that 15:36:39 q? 15:37:34 q? 15:37:57 Anssi: I propose we will have a joint discussion with the Privacy WG at TPAC to chart the path forward for new deliverables 15:38:08 Subtopic: TPAC F2F 15:38:18 q+ 15:38:26 Anssi: I want to use this opportunity to gauge F2F issue priorities, cross-group coordination and breakout interest 15:38:38 ... we have roughly 4 hours of F2F time for WebMCP specifically and 1 hour for built-in AI APIs 15:39:04 ... first, about F2F issues priorities 15:39:14 ... I'm suggesting the group to assign labels to issues they think are worth discussing during the F2F meeting, editors but anyone else as well: 15:39:21 -> https://github.com/webmachinelearning/webmcp/issues 15:39:48 Anssi: and respectively for built-in AI APIs 15:39:55 Anssi: please go to issues of your favorite repositories, use the existing labels or mint new labels to help indicate F2F discussion priorities and interest, or add a comment if you can't add labels 15:40:01 ... alternatively you can add comments to our meeting issue: 15:40:08 -> https://github.com/webmachinelearning/meetings/issues/39 15:40:09 https://github.com/webmachinelearning/meetings/issues/39 -> Issue 39 WebML WG/CG F2F Agenda - TPAC 2026 (Dublin, Ireland) (by anssiko) 15:40:45 Anssi: the group is executing like a well-oiled machine, so I don't want to freeze our agenda too early but consider dynamic adjustments based on ongoing discussions and emerging priorities 15:41:17 q? 15:41:19 ack domfarolino 15:41:44 Dominic: how many group we are meeting at TPAC? 15:42:03 ... we have joint meetings with horizontals, and breakout with Web Extensions WG 15:43:46 ... I can propose WebMCP breakout session, if that is helpful 15:45:04 Dominic: perhaps time on technical discussion in this CG is adequate 15:45:10 -> https://github.com/w3c/tpac2026-breakouts/ 15:45:33 q? 15:45:39 Subtopic: Cross-document tool response mechanism 15:45:46 Anssi: issue #135 and PR #327 15:45:46 https://github.com/webmachinelearning/webmcp/pull/327 -> PR 327 Add explainer for WebMCP continuations. (by markafoltz) 15:45:46 https://github.com/webmachinelearning/webmcp/issues/135 -> Issue 135 Cross-document tool response mechanism (by domfarolino) [Agenda+] [declarative] 15:45:58 ... prior, we resolved this issue to explore a solution for the declarative API 15:46:12 ... MarkF has prepared an explainer for a solution for the imperative API 15:46:19 ... building upon a proposed new higher-level abstraction based on token creation and token redemption, called Continuation Tokens 15:46:23 ... the explainer PR is out for review: 15:46:27 -> https://github.com/webmachinelearning/webmcp/blob/cont-explainer/continuations-explainer.md 15:47:10 MarkF: problem statement is that from the agent's point of view, the tool can span multiple documents 15:47:33 ... navigating to another document, or tool does work in one frame and wants to hand off work to another frame within the same frame tree 15:47:42 ... currently a tool cannot span multiple documents 15:48:30 ... this proposes a mechanism where an agent can request a token and pass the token that can be redeemed and continue working 15:48:44 ... enables chain of tool execution between documents 15:49:15 ... initial proposal enables tokens to be produced and consumed in same browsing context, i.e. popup or another tab cannot consume tokens from another tab 15:49:59 ... agents can see websites as a single tool, example is a checkout flow 15:50:11 q? 15:50:17 Anssi: any open questions? 15:50:32 MarkF: how to communicate the token across documents 15:50:57 ... API shape in general, if you have app spanning pages, and you want to implement tools across them 15:50:58 q? 15:51:48 Dominic: we would welcome web platform feedback, also agent implementer feedback, Copilot folks feedback, OpenAI feedback, this is important for agents that drive WebMCP tools 15:52:10 ... native agents with access to browser internals, feedback from them is important 15:52:11 q? 15:53:06 Sushanth: in our prototyping we handles passing between documents, so that tool call had an additional value in JSON 15:53:31 ... would like to understand why you are not considering passing state across document navigation, maybe for security reasons? 15:53:57 MarkF: browser state willing to provide is original input, actuation of the tool, if there's state beyond it, it is up to the application 15:54:09 ... we allow tool continuations to see partial outputs 15:54:10 q? 15:54:37 Dominic: did Microsoft's prototype return a specific value? 15:54:54 Sushanth: yes, on the orchestration value, after the page navigation 15:55:21 MarkF: earlier proposal had this re-entrant tool calling, we felt better to have this design we're proposing now 15:55:22 q? 15:55:40 Sushanth: passing state between tool location was part of the protocol in our implementation 15:55:53 ... in your proposal you make it an app problem, not WebMCP problem 15:56:09 Dominic: can consider partial state, we want to see partial state exposed to the model 15:56:28 ... we need to figure out what the entry point is, do we force the developer to provide continuation for the previous tool 15:56:42 ... a tool would not need to be registered on the target page 15:56:46 Sam has joined #webmachinelearning 15:56:50 q? 15:57:04 Sushanth: how does it know what methods to call to continue with the tool? 15:57:16 MarF: Sushanth, please provide feedback in the issue 15:57:16 q? 15:58:11 +1 15:58:18 RESOLUTION: The group explores the Continuation Tokens design outlined in the explainer and welcomes further feedback. (issue #135 and PR #327) 15:58:53 RRSAgent, draft minutes 15:58:55 I have made the request to generate https://www.w3.org/2026/10/01-webmachinelearning-minutes.html anssik 15:59:12 Subtopic: Signaling tool stability to agents 15:59:17 Anssi: issue #328 15:59:17 https://github.com/webmachinelearning/webmcp/issues/328 -> Issue 328 Signaling tool stability to agents (by domfarolino) [Agenda+] [needs-agent-feedback] 15:59:27 Dominic: would love Microsoft's feedback 15:59:50 Sam: Microsoft has interest in this, there is need for this and I can take another look to provide additional feedback 15:59:51 q? 16:00:06 Topic: Prompt API 16:00:10 gb, this is webmachinelearning/prompt-api 16:00:10 anssik, OK. 16:00:13 Subtopic: Thinking mode 16:00:18 Anssi: PR #217 16:00:19 https://github.com/webmachinelearning/prompt-api/pull/217 -> MERGED PR 217 Add Thinking mode section to the Prompt API explainer (by isaacahouma) [Agenda+] 16:00:22 ... explainer available: 16:00:32 -> https://github.com/webmachinelearning/prompt-api#thinking-mode 16:01:07 Reilly: feedback requested for this feature 16:01:08 q? 16:01:15 RRSAgent, draft minutes 16:01:16 I have made the request to generate https://www.w3.org/2026/10/01-webmachinelearning-minutes.html anssik 16:02:55 Present+ Alex_Nahas 16:03:03 Present+ Ali_Spivak 16:03:10 Present+ Andrew_Nolan 16:03:16 Present+ Ben_Greenstein 16:03:37 RRSAgent, draft minutes 16:03:38 I have made the request to generate https://www.w3.org/2026/10/01-webmachinelearning-minutes.html anssik 16:03:52 Present+ Ehsan_Toreini 16:04:17 Present+ Johann_Hofmann 16:04:29 Present + Iris_Johnson 16:04:37 Present+ Iris_Johnson 16:04:44 s/Present + Iris_Johnson// 16:05:10 Present+ Mike_Wasserman 16:05:16 Present+ Reilly_Grant 16:05:24 Present+ Ryan_Roemer 16:05:35 Present+ Sam_Dallstream 16:05:50 Present+ Sushanth_Rajasankar 16:06:14 Present+ Victor_Huang 16:06:43 Present+ Francois_Daoust 16:06:59 Present- tidoust 16:07:06 RRSAgent, draft minutes 16:07:08 I have made the request to generate https://www.w3.org/2026/10/01-webmachinelearning-minutes.html anssik 16:11:15 s/be to/be that the 16:12:44 s/we've be happy/we'd be happy 16:16:03 s/someone in scope/somewhat in scope 16:17:14 s/many group/many groups 16:17:53 s/perhaps time/perhaps focusing 16:18:52 s/or tool/or when the tool 16:20:03 s/handles/handled 16:20:44 s/state willing/state we are willing 16:21:13 s/orchestration value/orchestration layer 16:22:10 s/MarF/MarkF 16:22:37 RRSAgent, draft minutes 16:22:38 I have made the request to generate https://www.w3.org/2026/10/01-webmachinelearning-minutes.html anssik 19:09:40 Zakim has left #webmachinelearning 21:28:18 croissant` has joined #webmachinelearning