fullstackchris

This comment makes no sense here. Did you read the article? The author built an entire SaaS app in a few days with an agent. That isn't "just search"

You're suggesting their arent any other authors who have taken over a decade to write a book? Prousts' In Search of Lost Time took 13 for example.

> deep in the weeds with AI already

I doubt even 10% have written a custom MCP tool... and probably some who don't even know what that means

There is a reason it is not even mentioned

This is incorrect. Almost all brokers nowadays allow for OAuth connections where you can stream any data you'd see in your broker platform to any consuming web application (given the consuming web application does all the paperwork, dev work, and so on)

Well, there are no consequences are there? Or at least no precident of consequences of such behaviour. My hope is that folks like this always lose out in the long run but I'm not so sure anymore...

Second this. Sick of seeing posts like this because correlation =/= causation, proved time and time again. It's just too easy to 'relate' these two things and leads to lazy writing / persisting this narrative which has in no way been proven to be true yet.

IMO the underlying cause has much more to do with a hiring cycle issue: the boom of the low-interest / free money / I-don't-need-to-pay-for-an-office covid years is now leading to the relative hiring "bust" (even though it's not really a bust, unemployment is at 4.2%, certainly nothing out of the ordinary for the US)

still hacking away at codevideo - basically event sourcing for the IDE https://codevideo.io

its good enough for me that ive started using it for my MCP masterclass videos / code export / transcript https://mcpmasterclass.com

clerk is an auth library - and finally one that doesnt require dozens of lines to do things like, i dont know, check if the user is logged in

and wild... you used gemini to process a screenshot to find the website for a 5 letter word library?

Some of the best software learning I ever had when I was starting out was following along with video courses and writing the code line by line along with the instructor... or does this not count as "consumption"?

Gonna be a bit blunt here and ask why hooking up an agentic CLI tool to one or more other software tool(s) is the top post on HN right now... sure, some of these ideas are interesting but at the end of the day literally all of them have been explored / revisited by various MCP tools (or can be done more or less in scripted / hacked ways as the author shows here)

I don't know, just feels like a weird community response to something that is the equivalent to me of bash piping...

Yeah I was done at "What happened here was more than just code..." -_-

this behaviour is literally removable with proper prompting.

this is a strawmans argument... of whatever your are arguing

also, i will say, (if we can trust the findings in these notes are relatively accurate of the real implementation) is a PERFECT example of the real level of complexity used in cutting edge configuration of using LLM... its not just some complex fancy prompt you give to a model in a chat window... there is so much important stuff happening behind the scenes... though i suppose the people who complain about LLMs hallucinating / screwing up havent tried claude code or any agentic work flows - or, it could be their architecture / code is so poorly written and poorly organized that even the LLM itself struggles to modify it properly

interesting... the analysis finds that the MCP supports websockets as a transport... when there is big drama going on right now that anthropic said "they will never support that", folks hating SSE, and so on and so forth

ITT: access tokens really work!

Really a waste of time topic but "interesting" I suppose for people who don't understand the tools themselves

CodeVideo - https://github.com/codevideo

Basically, event sourcing for the IDE - build entire software lessons and course videos without actually recording anything.

You can get up and running pretty quickly at our studio https://studio.codevideo.io

Or locally with the CLI tool: https://github.com/codevideo/codevideo-cli

Great article, yes, its the classic "I want to not work AND keep my current lifestyle" I dont know if it was social media, bitterness at boomers, or whatever, that made most zoomers behold this pipe dream.

Loved the Massena example, but I went to school near there (Clarkson) and the winters are pretty rough. Summer / fall is beautiful though!

In that case grep is fine, but if I have a specific artifact I need to transport from one function to another, I'll need some sort of background set / get.

from my experience its about the speed of a very competant human - one of my favorite custom tools ive written is just access to a series of bash commands - havent tested with others but claude very quickly browses through files, reads them, and so on to do whatever it was you prompted. But even then it is all contextual - for example, I had to remove 'find' because as one would expect, running 'find' against a huge directory set is very slow!

This is exactly what I've encountered, at least with Claude, it writes out huge artifacts (static ones retrieved from the file system or wherever) character for character - What I'm going to try this weekend is just integrating a redis cache or sqlite into the MCP tool calls, so claude doesnt have to write everything out character per character... no idea if it will work as expected...

also looking into "fire and forget" tools, to see even if that is possible

Was gonna say, I thought this already happened last year...

Reading these threads its clear to me people are so cooked and no longer understand (or perhaps never did) understand the simple process of how source code is shared, built, and merged together with multiple editors has ever worked

the stock market looked like that chart in 2008 and look where we are now

unless LLMs can be instantly trained on all new software frameworks and languages that come out, im not worried stackoverflow will still have a place

Somehow it's ironic that this post is about AI replacing jobs and yet when I click the "accept all" option for cookies, the page reloads and shows the banner again :)

Where are the all knowing AI bots who are going to fix this?

> you'll have customers that demand you rewrite your app just because they gave you $9

I've also seen this first hand. Customer emails me 3 times within 3 days and one of those emails included him stating that "time is precious". He subscribed on the $7 / month tier... and then cancelled the next day anyway because he didnt even read what the product did... sigh

Well... if the author wants websockets, why doesn't the author just do it? They state clearly in pull request the community is totally free to make their own.

My guess is one will crop up within the next few months...

Speaking from the TypeScript side of things...

I will say the documentation is indeed garbage - it still includes code snippets of APIs / source code examples that don't exist anymore. Also, the choice to use zod types is also in my opinion, way over the top... the fact that I need to import a third party library to write a MCP server is wild - when you're already writing in a typed language (TypeScript)... (and yes I know the other advantages zod provides)

Otherwise it's simple enough to get started, if just tinkering around.

Here's a custom MCP tool I use to run commands and parse stdout / stderr all the time:

    try {
        const execPromise = promisify(exec);
        const { stdout, stderr } = await execPromise(command);
        if (stderr) {
            return {
                content: [{
                    type: "text",
                    text: `Error: ${stderr}`
                }],
                isError: true
            };
        }
        return {
            content: [{
                type: "text",
                text: stdout
            }],
            isError: false
        };
    } catch (error: any) {
        return {
            content: [{
                type: "text",
                text: `Error executing command: ${error.message}`
            }],
            isError: true
        };
    }

Yeah, if you want to be super technical, it's Node that does the actual command running, but in my opinion, that's as good as saying the MCP client is...

> early mistakes of Python?

Python.

Once I gave claude read only access to the command line and also my local repos, i found that was enough to have it work quite well... I start to wonder if all this will boil down to simple understanding of some sort of "semantic laws" still fuzzily described... I gotta read chomsky...