Drop the stale native-agents note from the security questionnaire - #269
Drop the stale native-agents note from the security questionnaire#269lazizbekravshanov wants to merge 2 commits into
Conversation
There was a problem hiding this comment.
Pull request overview
Updates the security/privacy questionnaire to align with the current WebMCP spec wording around exposedTo, removing an outdated aside about native-agents from earlier design discussions.
Changes:
- Removes the stale parenthetical about
native-agentsin theexposedTodescription. - Keeps the questionnaire statement focused on the current spec behavior: origins control tool discovery.
馃挕 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.
| The feature is gated by the [`"tools"`](https://webmachinelearning.github.io/webmcp/#permissiondef-tools) permission policy. It is allowed in top-level documents and same-origin descendants by default; The permission policy can be used to allow it in cross-origin iframes and/or to disallow it in same-origin frames. | ||
|
|
||
| Additionally, tools can specify [`exposedTo`](https://webmachinelearning.github.io/webmcp/#dom-modelcontextregistertooloptions-exposedto) to control which origins (or `native-agents`, name to be bikeshed per [#179](https://github.com/webmachinelearning/webmcp/pull/179)) can discover them. | ||
| Additionally, tools can specify [`exposedTo`](https://webmachinelearning.github.io/webmcp/#dom-modelcontextregistertooloptions-exposedto) to control which origins can discover them. |
There was a problem hiding this comment.
Good catch, the description overclaimed. The README's built in agent exposure section does discuss a proposed native-agent keyword and that is the live home for the idea. What is stale is the questionnaire citing merged #179 as a pending bikeshed. I updated the PR description to say this precisely.
There was a problem hiding this comment.
I agree the reference to #179 is stale, but I think the underlying intent is still there to support something like native-agent so tool authors in cross-origin iframes can explicitly control exposure to the browser's built-in agent.
If we do want to land this PR, could we update the reference here to point to the open design question in the README instead of removing the concept entirely?
|
anssiko marked as non substantive for IPR from ash-nazg. |
The security questionnaire says
exposedTocontrols which origins (ornative-agents, name to be bikeshed per #179) can discover tools. That parenthetical is stale: #179 merged in May 2026, the spec today restrictsexposedToto potentially trustworthy origins only, and thenative-agentstoken does not exist in the spec.A related idea is still live: the README's built in agent exposure section discusses a proposed
native-agentkeyword forexposedTo. That section is the right home for the open design question. What is stale here is the questionnaire pointing at a merged PR as a pending bikeshed, so this drops the parenthetical to match the current spec.