15:52:55 RRSAgent has joined #webmachinelearning 15:52:59 logging to https://www.w3.org/2026/02/19-webmachinelearning-irc 15:52:59 Meeting: WebML CG Teleconference – 19 February 2026 15:52:59 RRSAgent, make logs Public 15:53:00 please title this meeting ("meeting: ..."), anssik 15:53:01 Chair: Anssi 15:53:06 Agenda: https://github.com/webmachinelearning/meetings/blob/main/telcons/2026-02-19-cg-agenda.md 15:53:13 Scribe: Anssi 15:53:17 scribeNick: anssik 15:53:30 Present+ Anssi_Kostiainen 15:53:35 RRSAgent, draft minutes 15:53:36 I have made the request to generate https://www.w3.org/2026/02/19-webmachinelearning-minutes.html anssik 15:59:10 BenGreenstein has joined #webmachinelearning 15:59:27 Present+ Ben_Greenstein 16:00:25 Present+ Harsh_Varshney 16:00:35 brwalder has joined #webmachinelearning 16:00:45 Present+ Christian_Liebel 16:01:48 chrishtr has joined #webmachinelearning 16:01:49 Present+ Alex_Nahas 16:01:54 Present+ Brandon_Walderman 16:01:57 present+ 16:02:06 Present+ Chris_Harrelson 16:02:10 Present+ Dominic_Farolino 16:02:24 Present+ Iris_Johanson 16:02:34 Present+ Kasper_Kulikowski 16:02:42 Present+ Reilly_Grant 16:02:46 Present+ Vasilii_Trofimchuk 16:02:50 Present+ Victor_Huang 16:02:56 qcomp has joined #webmachinelearning 16:03:05 Present+ Jason_McGhee 16:03:06 jason has joined #webmachinelearning 16:03:08 Kasper_Kulikowski has joined #webmachinelearning 16:03:12 domfarolino has joined #webmachinelearning 16:03:16 thelounge has joined #webmachinelearning 16:03:29 Present+ Ben_VanderSloot 16:03:42 AlexN has joined #webmachinelearning 16:03:54 Ehsan has joined #webmachinelearning 16:03:54 Fabio has joined #webmachinelearning 16:03:55 mtavenrath has joined #webmachinelearning 16:04:15 Harsh has joined #webmachinelearning 16:04:15 Iris has joined #webmachinelearning 16:04:20 ygupta81 has joined #webmachinelearning 16:04:20 Present+ Achim_Schlosser 16:04:24 Victor has joined #webmachinelearning 16:04:30 lgombos has joined #webmachinelearning 16:04:31 aschlosser has joined #webmachinelearning 16:04:32 Present+ Ehsan_Toreini 16:04:39 Present+ Laszlo_Gombos 16:04:39 Present+ Fabio_Bernardon 16:04:50 Present+ Kryspin_Ziemski 16:05:05 Present+ Markus_Tavenrath 16:05:37 RRSAgent, draft minutes 16:05:39 I have made the request to generate https://www.w3.org/2026/02/19-webmachinelearning-minutes.html anssik 16:05:55 johannhof has joined #webmachinelearning 16:06:22 Present+ Johann_Hofmann 16:06:26 RRSAgent, draft minutes 16:06:27 I have made the request to generate https://www.w3.org/2026/02/19-webmachinelearning-minutes.html anssik 16:06:42 Anssi: to start, we will welcome our latest new participants: 16:06:50 ... Ilya Grigorik and Ates Goral from Shopify 16:07:00 ... Julia Pagnucco from Google 16:07:26 ... Qi Pan from Xiaoyuzhou 16:07:45 ... Achim Schlosser from Bertelsmann SE & Co. KGaA 16:08:04 ... Tania Millan, David Crowe as an individual contributor 16:08:09 ... welcome all to the WebML Community Group! 16:08:20 Topic: WebMCP 16:08:36 gb, this is webmachinelearning/webmcp 16:08:36 anssik, OK. 16:08:51 Subtopic: State of the Union 16:09:01 bvandersloot has joined #webmachinelearning 16:09:12 -> WebMCP - Draft Community Group Report https://webmachinelearning.github.io/webmcp/ 16:09:34 Anssi: last time we welcomed new WebMCP spec editors Brandon, Khushal and Dominic and discussed practicalities of spec editing: 16:09:39 -> Prior meeting minutes https://www.w3.org/2026/02/05-webmachinelearning-minutes.html#6409 16:10:09 Anssi: we're now off to a great start with the spec and have diverse talent in this group that keeps on growing 16:10:21 ... to that end, I want to make sure we make the best use of our talent pool 16:10:38 ... and ensure everyone's contributions are well positioned 16:11:01 ... first, I want to establish that all group participants, not just editors, are welcome to take an active role is advancing issue discussions, and proposing PRs 16:11:10 ... no PR is too small! 16:11:36 ... the editors will help review the proposals shared in GH issues and PRs 16:11:45 ... if there are questions on contribution logistics, I'm always happy to help 16:12:37 ... I have invited the broader web community to join us and I'm happy to see WebMCP experiments pop up that bring this technology to life and inform the spec development and its priorities 16:13:10 ... the early seeds of the WebMCP spec draft were planted by prototyping activity by Alex and Jason, now WebMCP has advanced into browser vendors' experimental implementations 16:15:29 Achim: I'm working on data standards at Bertelsmann, looking the impact of business models on media consumption and change the AI brings 16:15:39 chrome canary releasing the webmcp implementation + going viral on various platforms caused massive influx in interest as you can see with the star history of the repo. 500 -> 1.4k over a few days (https://www.star-history.com/#webmachinelearning/webmcp&type=date&legend=top-left) 16:15:59 Anssi: to that end, I'd like to share an update on the implementation front: 16:16:08 Anssi: an early preview of WebMCP is now available behind a flag in Chrome 146 16:16:14 Anssi: here's how to test this early experimental WebMCP today: 16:16:17 ... 1) get Chrome Canary 146+ 16:16:22 ... 2) enable chrome://flags/#enable-webmcp-testing 16:16:26 q+ 16:16:29 ... 3) try experimental WebMCP demos, e.g. those shared in issue #90: 16:16:30 https://github.com/webmachinelearning/webmcp/issues/90 -> Issue 90 [Improvement]: List of sites supporting WebMCP (by javixeneize) 16:16:44 -> Experimental WebMCP utils and demos by GoogleChromeLabs https://github.com/GoogleChromeLabs/webmcp-tools/blob/main/AWESOME_WEBMCP.md 16:16:54 -> WebMCP bridge by Hemanth https://h3manth.com/ai/webmcp/ 16:17:07 q- 16:17:16 -> Model Context Tool Inspector by Francois https://github.com/beaufortfrancois/model-context-tool-inspector 16:17:26 Anssi: WebMCP bridge is an interesting experiment that lets you connect to any remote MCP server and exposes its tools to WebMCP API 16:18:50 Dominic: Early Preview Program is available to gather feedback from early adopters re early implementation 16:19:47 Anssi: if we look at the state of the spec 16:20:06 ... besides the algorithmic prose, there's a lot of informative text that welcomes contributions from all group participants: 16:20:08 qcomp has joined #webmachinelearning 16:20:32 ... for example, introduction, terminology, security, privacy and accessibility sections, also sample code snippets 16:20:59 q+ 16:21:06 ... below the IDL blocks we have non-normative green "for web developers" descriptions that benefit from review and comments from web developers who are the users of the API 16:21:22 ... the normative descriptions are guidelines for browser implementers that ensure interoperability 16:21:27 ack chrishtr 16:21:56 Chris: considering the excitement could be make this meeting weekly? 16:22:15 +1 16:22:36 Is there appetite for recording calls for folks that cannot attend? 16:23:34 Victor has joined #webmachinelearning 16:23:41 q? 16:23:47 Subtopic: Updated declarative proposal 16:24:01 Anssi: the group agreed to evolve the declarative proposal alongside the imperative WebMCP 16:24:19 ... an updated declarative proposal is discussed in issue #22 and the proposal itself is in PR #76, superseding PR #26 16:24:20 https://github.com/webmachinelearning/webmcp/pull/76 -> Pull Request 76 Declarative API Explainer (by domfarolino) [Agenda+] 16:24:20 https://github.com/webmachinelearning/webmcp/issues/22 -> Issue 22 Declarative API Equivalent (by EisenbergEffect) [declarative] 16:24:20 https://github.com/webmachinelearning/webmcp/pull/26 -> CLOSED Pull Request 26 Declarative proposal to expose tools via HTML (by MiguelsPizza) 16:24:42 ... PR #76 is currently marked as draft 16:24:47 ... Dominic, would you like to introduce the new declarative explainer and the remaining TODOs? 16:25:03 Dominic: the PR is now ready for rewiew 16:25:15 ... it's in a good shape as an initial explainer 16:25:48 ... on top of this PR, for TODOs, we want to spec input schema details 16:25:52 -> WebMCP declarative API (PR preview) https://github.com/webmachinelearning/webmcp/blob/declarative-api-explainer/declarative-api-explainer.md 16:26:07 Dominic: TODOs are not blocking the initial PR from landing 16:26:09 q? 16:26:11 q+ 16:27:15 Dominic: I'm also working on a bigger spec change to improve internal structures on how to register tools, with reusability in mind 16:27:31 q? 16:27:40 ack bvandersloot 16:28:20 BenV: I think the explainer is in a good share, a few questions I'll share in the PR soon 16:28:21 q? 16:28:49 q? 16:29:39 Anssi: in issue #22 I see the community is developing more prototypes 16:29:40 https://github.com/webmachinelearning/webmcp/issues/22 -> Issue 22 Declarative API Equivalent (by EisenbergEffect) [declarative] 16:29:57 Subtopic: Agent allowlist 16:30:05 Anssi: issue comment https://github.com/webmachinelearning/webmcp/issues/51#issuecomment-3893594368 16:30:07 https://github.com/webmachinelearning/webmcp/issues/51 -> Issue 51 Define the API for in-page Agents to use a site's declared tools (by khushalsagar) [Agenda+] 16:30:19 ... Alex brought up a proposal for the "agent allowlist" to give more agency to the user and web author 16:30:25 ... the motivating use case is from an enterprise perspective 16:30:47 ... an enterprise may have developed its own internal browser extension for its userbase and web apps that define tools via WebMCP 16:30:58 ... and the same enterprise wants to allow its own extension to call these WebMCP tools, but does not want to expose these same tools to 1P agents built into the browser 16:31:08 ... I believe there may be e.g. legal or compliance reasons that vary by country that would require such a feature 16:31:23 ... I could envision finance or health sector may have stricter requirements in this regard 16:31:50 ... would the group support further work on this feature, starting with documenting this use case to understand the requirements? 16:31:52 q? 16:32:30 q+ 16:32:35 q+ 16:32:38 Alex: that was spot on, companies have comliance reasons for this feature to exists 16:32:41 q? 16:33:00 Victor has joined #webmachinelearning 16:33:00 qcomp has joined #webmachinelearning 16:33:00 bvandersloot has joined #webmachinelearning 16:33:00 johannhof has joined #webmachinelearning 16:33:00 aschlosser has joined #webmachinelearning 16:33:00 ygupta81 has joined #webmachinelearning 16:33:00 Iris has joined #webmachinelearning 16:33:00 Harsh has joined #webmachinelearning 16:33:00 mtavenrath has joined #webmachinelearning 16:33:00 Fabio has joined #webmachinelearning 16:33:00 Ehsan has joined #webmachinelearning 16:33:00 AlexN has joined #webmachinelearning 16:33:00 jg has joined #webmachinelearning 16:33:00 domfarolino has joined #webmachinelearning 16:33:00 Kasper_Kulikowski has joined #webmachinelearning 16:33:00 jason has joined #webmachinelearning 16:33:00 brwalder has joined #webmachinelearning 16:33:00 BenGreenstein has joined #webmachinelearning 16:33:00 reillyg has joined #webmachinelearning 16:33:02 ack reillyg 16:33:40 Reilly: I'm only familiar with Chrome enterprise features, e.g. limiting copy-pasting between sites 16:34:29 ... it could be an enterprise policy setting in browser 16:34:39 Victor has joined #webmachinelearning 16:34:39 qcomp has joined #webmachinelearning 16:34:39 bvandersloot has joined #webmachinelearning 16:34:39 johannhof has joined #webmachinelearning 16:34:39 aschlosser has joined #webmachinelearning 16:34:39 ygupta81 has joined #webmachinelearning 16:34:39 Iris has joined #webmachinelearning 16:34:39 Harsh has joined #webmachinelearning 16:34:39 mtavenrath has joined #webmachinelearning 16:34:39 Fabio has joined #webmachinelearning 16:34:39 Ehsan has joined #webmachinelearning 16:34:39 AlexN has joined #webmachinelearning 16:34:39 jg has joined #webmachinelearning 16:34:39 domfarolino has joined #webmachinelearning 16:34:39 Kasper_Kulikowski has joined #webmachinelearning 16:34:39 jason has joined #webmachinelearning 16:34:39 brwalder has joined #webmachinelearning 16:34:39 BenGreenstein has joined #webmachinelearning 16:34:39 reillyg has joined #webmachinelearning 16:34:39 Is there some duplication with https://github.com/webmachinelearning/webmcp/issues/96? 16:34:40 https://github.com/webmachinelearning/webmcp/issues/96 -> Issue 96 Proposal: [Section 3] Address agent-to-tool trust — identity, scoped permissions, and delegation context (by davidcrowe) 16:35:09 Alex: iframe in a page vs. libraries built by the community 16:35:14 ack dom 16:35:52 Chrome WebMCP Early Preview: https://docs.google.com/document/d/1rtU1fRPS0bMqd9abMG_hc6K9OAI6soUy3Kh00toAgyk/edit?tab=t.0 16:35:56 Dominic: how enforceable this feature would be? 16:36:25 q+ 16:36:31 ... it is all first-party script, we can't use same-origin policy restrictions, for iframe case it makes sense 16:36:34 q? 16:37:05 ack johannhof 16:38:12 Johann: I agree we should discuss this feature in a GH issue to understand the requirements for filtering by agent 16:38:57 q? 16:40:44 Dominic: this could be discussed in the agent-client issue also 16:41:39 Anssi: any objections to document UCs and reqs? 16:41:45 [none heard] 16:41:50 RESOLUTION: Open a dedicated issue to document "agent allowlist" use cases and requirements 16:42:00 Subtopic: Stream tool arguments 16:42:09 Anssi: issue #82 16:42:10 https://github.com/webmachinelearning/webmcp/issues/82 -> Issue 82 Proposal: Stream Tool Arguments (by MiguelsPizza) [Agenda+] 16:42:16 ... another proposal from Alex to improve both DX and UX 16:42:28 ... this proposal suggests a mechanism to allow streaming of chunks into page UI as the model generates them 16:42:38 ... Alex explains that currently execute method fires only after the full arguments JSON is parsed 16:42:49 ... this is problematic with large args such as email bodies and code 16:42:54 ... an example of streaming tool args is CopilotKit Research Helper: 16:42:58 -> https://www.copilotkit.ai/examples/canvas-research 16:43:43 Alex: I think this closes the gap between the userland and WebMCP, you may generate large JSON with a loading spiller, this makes the document look more live 16:43:45 q? 16:44:29 Anssi: any comments or questions? 16:44:51 Dominic: would the input schema not include primitives? 16:45:07 ... maybe we want to augment input schema to allow streaming rather than allow streaming JSON? 16:45:42 Alex: have an issue open for browser schema validation 16:46:03 q? 16:46:03 https://github.com/webmachinelearning/webmcp/issues/92 16:46:05 https://github.com/webmachinelearning/webmcp/issues/92 -> Issue 92 Who owns the validation layer? (by MiguelsPizza) 16:46:20 q? 16:46:39 Present+ Winston_Chen 16:47:17 Subtopic: WebExtensions API to enumerate and invoke tools 16:47:23 Anssi: issue #74 16:47:24 https://github.com/webmachinelearning/webmcp/issues/74 -> Issue 74 WebExtensions API to enumerate and invoke tools (by reillyeon) [Agenda+] 16:47:28 ... a proposal from Reilly 16:47:42 ... this issue is spun off from issue #23 by Jason that discussed three different categories of agents and how they might interface with WebMCP 16:47:43 https://github.com/webmachinelearning/webmcp/issues/23 -> Issue 23 External MCP Client API / Permissions (by jasonjmcghee) 16:47:56 ... 1) 1P in-browser agents, use browser-specific internal APIs 16:48:14 ... 2) 3P in-browser agents ~ browser extensions, use APIs defined by the WebExtensions group 16:48:29 ... 3) external desktop agents, use OS provided native APIs 16:48:47 q+ 16:48:59 ... Reilly proposes an approach to address the second case by collaborating with the WebExtensions group to define an API for use by browser extensions to enumerate an invoke tools defined with WebMCP 16:49:03 ... I see support in the issue discussion for the proposed approach 16:49:06 ack reillyg 16:49:51 Reilly: motivation is to make sure folks working on WebMCP are aware of WebExtensions 16:50:40 ... it's be nice to have a proper WebExtensions API to expose these features to browser extensions 16:50:41 q? 16:51:30 Dominic: does the proposal in issue #51 satisfy this request? 16:51:31 https://github.com/webmachinelearning/webmcp/issues/51 -> Issue 51 Define the API for in-page Agents to use a site's declared tools (by khushalsagar) [Agenda+] 16:51:51 q+ 16:51:53 q+ 16:52:04 Dominic: would like to hear from extension folks whether that covers the use case 16:52:07 ack reillyg 16:52:47 q+ 16:53:12 Reilly: extensions script calling a page script is not possible cleanly, this should be easy to do 16:53:13 q? 16:53:29 thats good to know - web ext API and normal API would do the same under the hood but be exposed in different contexts 16:53:48 Dominic: what's the complexity there, you can to inject a script, rather than call an explicit API? 16:54:52 q+ 16:55:37 Reilly: the complexity is injecting a script into the page that can interact with a DOM, communication based on postMessage, also makes permissions more complex 16:55:57 q? 16:56:07 q? 16:56:35 ack Victor 16:57:10 ack brwalder 16:57:25 ack AlexN 16:57:31 ack AlexN 16:58:02 Subtopic: WebMCP <-> MCP Apps bridge 16:58:07 Anssi: issue #57 16:58:08 https://github.com/webmachinelearning/webmcp/issues/57 -> Issue 57 Allow iframes to declare tools (by khushalsagar) [Agenda+] 16:58:17 Anssi: I think this issue relates on a high-level with the previous topic 16:58:26 ... in the previous topic we discussed how to bridge WebMCP with WebExtensions 16:58:38 ... in this topic I'd like to continue discussion on how WebMCP bridges with MCP Apps 16:58:41 -> Prior discussion https://www.w3.org/2026/02/05-webmachinelearning-minutes.html#e509 16:58:46 -> MCP Apps docs https://modelcontextprotocol.io/docs/extensions/apps 16:58:56 Anssi: for context setting, we recall MCP Apps run in a sandboxed iframe 16:59:05 ... do MCP Apps have specific requirements that differ from regular web content running in an iframe? 16:59:21 ... does the group consider MCP Apps a specific case of iframed web content, or an extension to it? 16:59:28 q? 16:59:43 q? 17:00:18 (re meeting minutes, anssi, do you have access to zoom captions/ full transcript? it might allow you to recover what folks said during the meeting) 17:00:23 Kryspin: issue #51 addresses one aspect of MCP Apps bridge, MCP Apps need to know what 3P agents are doing 17:00:24 https://github.com/webmachinelearning/webmcp/issues/51 -> Issue 51 Define the API for in-page Agents to use a site's declared tools (by khushalsagar) [Agenda+] 17:00:59 q? 17:02:28 q+ 17:02:33 Kryspin: for MCP Apps, the only thing is there's concurrency, listener to know there's tool execution and what is the state of execution, two authorities, MCP Agent and what's driving the MCP Apps 17:02:43 RRSAgent, draft minutes 17:02:44 I have made the request to generate https://www.w3.org/2026/02/19-webmachinelearning-minutes.html anssik 17:03:53 ack AlexN 17:04:11 Alex: this is something that can probably be polyfilled 17:04:25 Dominic: we need to ensure MCP Apps are compatible with WebMCP 17:04:48 Kryspin: if we implement issue #51 that'll solve a lot of this things 17:04:48 q? 17:05:51 RRSAgent, draft minutes 17:05:52 I have made the request to generate https://www.w3.org/2026/02/19-webmachinelearning-minutes.html anssik 17:06:59 Present+ James_Garbutt 17:07:01 RRSAgent, draft minutes 17:07:02 I have made the request to generate https://www.w3.org/2026/02/19-webmachinelearning-minutes.html anssik 17:08:29 s/is advancing/in advancing 17:11:03 s/could be make/could we make 17:11:53 s/rewiew/review 17:12:29 s/good share/good shape 17:13:29 s/comliance/compliance 17:13:41 s/exists/exist 17:14:10 s/it could be/could this be 17:16:15 s/spiller/spinner 17:18:00 s/you can to inject/you can inject 17:25:44 RRSAgent, draft minutes 17:25:46 I have made the request to generate https://www.w3.org/2026/02/19-webmachinelearning-minutes.html anssik 17:27:12 i was able to do it on the call