I think some of the advanced features around sampling from the calling LLM could theoretically benefit from a bidirectional stream.
In practice, nobody uses those parts of the protocol (it was overdesigned and hardly any clients support it). The key thing MCP brings right now is a standardized way to discover & invoke tools. This would’ve worked equally well as a plain HTTP-based protocol (certainly for a v1) and it’d have made it 10x easier to implement.
Sampling is to my eyes a very promising aspect of the protocol. Maybe its implementation is lagging behind because it's too far from the previous mental model of tool use. I am also fine if the burden is on the client side if it enables a good DX on server side. In practice, there would be much more servers than clients.
> This would’ve worked equally well as a plain HTTP-based protocol
With plain HTTP you can quite easily "stream" both the request's and the response's body: that's a HTTP/1 feature called "chunking" (the message body is not just one byte array, it's "chunked" so that each chunk can be received in sequence). I really don't get why people think you need WS (or ffs SSE) for "streaming". I've implemented a chat using just good old HTTP/1.1 with chunking. It's actually a perfect use case, so it suits LLMs quite well.
Well, the point is to provide context, it's easier to do if server has state.
For example, you have a MCP client (let's say it's amazon q cli), a you have a MCP server for executing commands over ssh. If connection is maintained between MCP client and server, then MCP server can keep ssh connection alive.
Replace SSH server with anything else that has state - a browser for example (now your AI assistant also can have 500 open tabs)
I have my childrens accounts under my main account. If they watch a lot of certain types of videos, that will show up as a suggestion under my main account. I'm positive its not because they are using my account, I watch videos with them.
I also have a second account for watching more technical videos and that leaks into my main account suggestions.
Yes it's definitely affected by location (IP) and/or specific device metadata, casting videos from an incognito window affects future recommendations even though it warns about not being signed in to preserve history.
Poison YouTube recommendations at the nearest Starbucks! (Actually a marketing hack - encourage followers to log out and watch a channel's YouTube videos on public WiFi.)
if you leave the mouse hovering over a video and, if it autoplays (previews) an $amount of the video, it will be added to your view history and you will then be offered more of that "content".
I turn this "feature" off every time I open it, it tells me it has saved that configuration, but that is not the case.
like other poster commented, review your view history and delete anything you do not want to see or clear it out entirely. There is also option to "don't recommend channel" ([3x ...] beside the video)
JIRA is a tool, it doesn't do the work for you. If someone knows how to organize, plan and track a project really well, the tool used doesn't matter as much.
Similar experience but I was in the military and I didn't have a choice to say no. I have 5 fillings. They wanted to schedule 2 more but I was due to be reassigned overseas. They tried to delay my reassignment over it. Eventually, they cleared my paperwork but made it clear I was to see the dentist at my new base as soon as possible.
My new dentist says there is nothing wrong with my teeth whatsoever.
reply