feat: add MCP Chrome extension #325
Conversation
mxschmitt
force-pushed
the
extension-drafft
branch
2 times, most recently
from
4280890 to
dcce7ae
Compare
| * @typedef {{tabId: number}} DebuggerTarget | ||
| */ | ||
|
|
||
| function debugLog(...args) { |
There was a problem hiding this comment.
this looks like a debugging remnant. is there value in commiting it?
|
|
||
| mcpExtensionPage: async ({ mcpExtension, mcpHeadless }, use) => { | ||
| if (!mcpExtension) | ||
| return await use(undefined); |
There was a problem hiding this comment.
throw an error instead, then you don't have to assert() on it on the calling side
There was a problem hiding this comment.
startClient uses this fixture, so if we remove the undefined we would lie about the type and throw there? So not sure if that suggestion still applies.
mxschmitt
force-pushed
the
extension-drafft
branch
2 times, most recently
from
ed6a65c to
eaf5fa3
Compare
mxschmitt
force-pushed
the
extension-drafft
branch
from
53be395 to
7040a41
Compare
|
Let's structure your code a bit differently:
|
mxschmitt
force-pushed
the
extension-drafft
branch
from
7040a41 to
0698d71
Compare
| this.dispatch({ id: message.id, result: versionInfo }); | ||
| return; | ||
| } | ||
| if (message.method === 'Target.setAutoAttach' && !message.sessionId) { |
There was a problem hiding this comment.
If message.sessionId exists, I assume it will just forward the request to chrome.debugger.sendCommand, right? I had some issues in the past with Target.setAutoAttach requests on pages with service workers:
There was a problem hiding this comment.
And there was also issues with oopif that I had to address on Target.setAutoAttach:
No description provided.