WEBVTT 00:00:00.020 --> 00:00:04.200 Hello and welcome to Python Bytes, where we deliver Python news and headlines directly 00:00:04.760 --> 00:00:11.240 to your earbuds. This is episode 490, recorded July 28th. I'm Michael Kennedy. 00:00:11.700 --> 00:00:12.740 And I'm Calvin Hendryx-Parker. 00:00:13.020 --> 00:00:17.920 This episode is brought to you by XWeather. XWeather combines enterprise-grade weather 00:00:18.140 --> 00:00:23.860 intelligence with agent-ready APIs, natural language capabilities, and an MCP server so 00:00:23.940 --> 00:00:28.359 your agents can adapt workflows, automate responses, and make better decisions based 00:00:28.380 --> 00:00:29.300 on real-world conditions. 00:00:29.800 --> 00:00:31.400 I'll tell you more about them later in the show. 00:00:31.760 --> 00:00:34.980 Get started for free at pythonbytes.fm/xweather. 00:00:35.300 --> 00:00:38.820 If you want to follow us on social, then feel free to do so. 00:00:39.070 --> 00:00:39.920 We got a link. 00:00:40.240 --> 00:00:43.000 I have a little bit more to say about that at the end of the show. 00:00:44.480 --> 00:00:48.280 Got links in the show notes and sign up for the newsletter, all those things. 00:00:48.700 --> 00:00:56.100 Before we get into the first topic, Calvin, I'd like to just rep a little bit of the KTM. 00:00:56.860 --> 00:00:58.120 See, I got my KTM shirt on. 00:00:58.280 --> 00:00:58.900 Team Orange. 00:00:59.520 --> 00:00:59.920 Team Orange. 00:00:59.940 --> 00:01:01.420 And I know you're a fan of motorcycles as well. 00:01:01.840 --> 00:01:02.060 I am. 00:01:02.220 --> 00:01:06.900 This week, this weekend, we had Pro AMA Motocross at Washougal. 00:01:07.320 --> 00:01:07.760 Oh, fun. 00:01:07.860 --> 00:01:12.520 And I was there all day with some of my friends, and it was quite the epic event. 00:01:12.760 --> 00:01:15.320 So if you're around, they travel all over the US. 00:01:15.900 --> 00:01:17.660 If you're around one of them, go to them. 00:01:17.740 --> 00:01:19.360 They're really an interesting experience. 00:01:19.840 --> 00:01:23.340 Even if you're not super into motorcycles, just being in that atmosphere is really wild. 00:01:23.640 --> 00:01:25.000 So it's cool. 00:01:25.240 --> 00:01:25.340 Yeah. 00:01:25.520 --> 00:01:27.900 I'm hoping they bring World Superbike someday back to Indianapolis. 00:01:28.420 --> 00:01:34.240 yes that would be awesome because i'm team ducati yeah well don't let uh daytona and their their 00:01:34.420 --> 00:01:39.320 bike week have all the fun you know yeah exactly exactly well you know what is fun i know it's fun 00:01:39.400 --> 00:01:45.400 jango's fun Django is a lot of fun yeah and in thinking about some of the core features of Django 00:01:46.020 --> 00:01:51.459 julia evans brings us great posts from last week about some of the things that she's been enjoying 00:01:51.460 --> 00:01:58.480 about Django, but from a 2010 style website perspective. I thought this was a great post 00:01:58.780 --> 00:02:04.900 because it reinforces some of the joy of Django and some of the cool bits and pieces that maybe 00:02:05.140 --> 00:02:10.679 people have not forgotten about or just aren't using or haven't chosen Django because they 00:02:10.740 --> 00:02:15.979 thought maybe it was just too complicated or too big for their project and they went Flask or Fast 00:02:16.000 --> 00:02:22.000 API instead. But this is a good one that kind of goes over what it's like, you know, again, 00:02:22.310 --> 00:02:30.420 the retro feel or nostalgia for a 2010 website was very backend. And with server side rendered pages, 00:02:31.080 --> 00:02:35.640 this is kind of giving you that feel for that. But then it gets into things like query builders, 00:02:36.680 --> 00:02:42.300 defining custom query sets, making them chainable, which if you take a look, this is very readable. 00:02:42.700 --> 00:02:50.560 Like if you were to read this code as English, you read like, you know, events for tab with festivals is free and is outdoors. 00:02:51.210 --> 00:02:54.040 That just almost makes sense to the human brain. 00:02:54.190 --> 00:03:03.640 But then it translates into the Django ORM really, really cleanly because you can set up your own custom query sets and have their own custom methods on it. 00:03:03.990 --> 00:03:09.520 So you can build a site that's a joy to maintain and to query behind the scenes. 00:03:09.980 --> 00:03:14.700 Another thing that she highlights in the post is actually template filters and template filters are awesome. 00:03:14.950 --> 00:03:24.720 I think a lot of people will neglect to take this into account when they are building web pages is if you can use these quick little filters that you can apply inside your template. 00:03:25.150 --> 00:03:30.400 And it gives you a ton of like the batteries included experience that you've come to love about Python. 00:03:30.920 --> 00:03:32.000 You get that in Django. 00:03:32.340 --> 00:03:39.100 And I think if you go and look at the reference for the template filters, you'll see there's just tons of cool little things that save you a ton of time. 00:03:39.180 --> 00:03:40.740 And you don't have to reinvent these wheels. 00:03:41.200 --> 00:03:42.380 They're all ready for you to go. 00:03:42.820 --> 00:03:50.600 And then there's also, for example, the JSON script, which takes a Python dictionary and automatically converts it to JSON and inserts into the HTML script tag in a safe way. 00:03:50.940 --> 00:04:05.780 So it also helps you with things like cross-site scripting and template injection and all those kind of security problems that normally you'd have to consider can go mostly to the wayside because Django just takes care of it for you when you use some of those built-ins. 00:04:06.560 --> 00:04:10.480 Other kind of things that we mentioned here, I believe, is like migrations. 00:04:11.530 --> 00:04:16.140 So if you are building websites, don't forget, if you change your database, you've got to migrate the code. 00:04:16.579 --> 00:04:20.940 And if you change the code, you got to migrate your database to match for those various models. 00:04:21.480 --> 00:04:25.240 Again, something that a lot of Django nuts don't think too much about because it just happens out 00:04:25.260 --> 00:04:32.620 of the box. But this is a good one that just kind of reinforces that. Another nicety that some folks 00:04:32.760 --> 00:04:38.280 like is skipping inheritance for class-based views. You can prefer function-based views for sharing 00:04:38.500 --> 00:04:44.399 code and then using Django's own mixins and interfaces to simplify that building of static 00:04:44.420 --> 00:04:49.380 sites. What's also nice, if you basically combine this with something like HTMX, you now get a 00:04:49.600 --> 00:04:52.700 superpower, which is you don't have to build your site twice. I think I mentioned this last week 00:04:53.000 --> 00:04:58.580 on the show, is that you build once and you've got a beautiful feel of a single page web app 00:04:58.920 --> 00:05:04.860 or a reactive style design, but without having to build your models twice. There's also some 00:05:05.540 --> 00:05:09.659 performance surprises. I think if you check out the latest version of Django, you'll notice that 00:05:09.680 --> 00:05:11.980 there's been some considerable performance improvements. 00:05:12.680 --> 00:05:16.360 If you're considering, say, FastAPI over Django with Ninja, 00:05:16.910 --> 00:05:18.680 you should, and that's not mentioned in this article necessarily, 00:05:18.890 --> 00:05:20.740 but I would mention that you could check that out 00:05:20.860 --> 00:05:23.980 and get fairly similar performance in certain use cases 00:05:24.550 --> 00:05:26.000 without having to go the FastAPI route 00:05:26.000 --> 00:05:29.000 and still getting all the batteries included of Django itself. 00:05:29.460 --> 00:05:31.720 So that was the kind of wrap up of the post. 00:05:32.060 --> 00:05:34.360 I'm going to mention it because I think sometimes people just need to be, 00:05:34.630 --> 00:05:38.119 have their memory jogged for some of the nice bits 00:05:38.140 --> 00:05:41.280 that are still baked into the default experience of Django. 00:05:41.660 --> 00:05:42.380 Yeah, that's super cool. 00:05:42.560 --> 00:05:44.120 You think admin backend. 00:05:44.360 --> 00:05:44.880 Oh, yeah. 00:05:45.240 --> 00:05:45.800 You know. 00:05:46.260 --> 00:05:47.100 But it's way more. 00:05:47.380 --> 00:05:48.000 It's so much more. 00:05:48.000 --> 00:05:48.260 Yeah, exactly. 00:05:48.420 --> 00:05:48.800 That's my point. 00:05:48.920 --> 00:05:51.300 There's a lot of these sort of like one bit lower, 00:05:51.860 --> 00:05:53.040 but still really helpful things. 00:05:53.500 --> 00:05:56.160 I feel like Julia needs to put a blog roll on this. 00:05:56.360 --> 00:05:56.820 I mean, come on. 00:05:56.920 --> 00:05:59.280 It's got no credibility with me until there's a blog roll. 00:06:01.260 --> 00:06:02.700 And if you don't know what a blog roll is, 00:06:03.080 --> 00:06:05.560 probably good, but it's certainly in the vibe. 00:06:05.560 --> 00:06:06.280 For going old school. 00:06:06.520 --> 00:06:09.500 It's just funny that word nostalgia for 2010s. 00:06:10.900 --> 00:06:11.200 Exactly. 00:06:11.500 --> 00:06:14.040 Can you believe how it was back then? 00:06:14.110 --> 00:06:16.480 That doesn't sound old to me, unfortunately. 00:06:16.980 --> 00:06:17.480 I know. 00:06:17.880 --> 00:06:19.300 So, yeah, yeah. 00:06:19.360 --> 00:06:20.940 I got a little bit of real-time follow-up. 00:06:20.990 --> 00:06:24.160 I just, the latest Talk Python episode that just came out yesterday. 00:06:24.160 --> 00:06:24.400 Oh, yeah. 00:06:24.720 --> 00:06:24.820 Yes? 00:06:25.040 --> 00:06:25.260 Yeah, yeah. 00:06:25.560 --> 00:06:26.660 Was with Carlton Gibson. 00:06:27.340 --> 00:06:32.499 And it's a really fun episode about updates on Django's async story 00:06:32.520 --> 00:06:37.480 and rewriting all the documentation and positioning of async for Django, 00:06:37.760 --> 00:06:40.220 plus talking some Django 6, 6.1 things. 00:06:40.440 --> 00:06:43.460 So if you liked Calvin's bit, there's a whole hour of it right there. 00:06:43.860 --> 00:06:44.600 That was really fun. 00:06:44.740 --> 00:06:46.460 I have a little more on Django in the extras too. 00:06:46.720 --> 00:06:47.420 Awesome. Awesome. 00:06:47.680 --> 00:06:48.120 Well, okay. 00:06:48.500 --> 00:06:52.220 Well, you know, it's all fun when there's a vibe coding party 00:06:52.840 --> 00:06:54.140 until you got to clean up the mess. 00:06:54.490 --> 00:06:55.760 So this got sent in. 00:06:57.100 --> 00:07:00.240 It got sent in to us by Sam Lerner saying, 00:07:00.840 --> 00:07:04.620 hey, there's this really interesting article on the Financial Times. 00:07:04.880 --> 00:07:06.580 So thank you, Sam, for sending that in. 00:07:07.000 --> 00:07:08.320 And it's paywalled. 00:07:08.700 --> 00:07:10.380 So I'm going to give you the rundown. 00:07:10.610 --> 00:07:13.920 But it's not so much that you need to read it, which is what's interesting. 00:07:14.280 --> 00:07:15.920 It's that it is here at all. 00:07:16.400 --> 00:07:19.300 So Financial Times is for business people. 00:07:19.680 --> 00:07:21.740 Yeah, I was curious where this was going, Michael. 00:07:22.100 --> 00:07:22.480 Exactly. 00:07:22.620 --> 00:07:23.400 Why is this here? 00:07:23.610 --> 00:07:26.220 So what is this story about? 00:07:26.360 --> 00:07:30.120 Who cleans up after the Vibe Coding Party is the title of the article. 00:07:30.580 --> 00:07:34.160 And the byline is hidden behind a paywall even, which is interesting. 00:07:34.560 --> 00:07:35.440 I was writing this article. 00:07:35.640 --> 00:07:37.540 I'd be like, could you just put my byline under it? 00:07:37.800 --> 00:07:38.020 Anyway. 00:07:38.420 --> 00:07:39.160 So it goes through. 00:07:39.160 --> 00:07:40.840 I'm not speaking to journalists in these days anyway. 00:07:41.440 --> 00:07:41.940 I know. 00:07:41.940 --> 00:07:44.160 You're like, I've already got it hard enough. 00:07:44.180 --> 00:07:44.700 Are you serious? 00:07:46.160 --> 00:07:49.100 So let me give you a little bit of the rundown for this one. 00:07:50.060 --> 00:07:51.980 Because there's one, because you don't have access to it. 00:07:51.980 --> 00:07:56.860 But I think the reason Sam sent this in and the reason I think it's interesting as well 00:07:57.240 --> 00:07:58.780 is actually, no, sorry. 00:07:58.920 --> 00:08:02.380 The piece is by Sam Lerner and Dylan McConnell sent it in. 00:08:02.680 --> 00:08:05.280 I wrote those down, but I had crisscrossed them in my mind. 00:08:05.450 --> 00:08:06.420 So thanks for sending that in, Dylan. 00:08:06.800 --> 00:08:10.780 So the reason I think this is interesting is it tells the story, 00:08:11.300 --> 00:08:17.000 but to a much broader audience who is not able to discern the details that I'm about to give you. 00:08:17.110 --> 00:08:17.220 Okay. 00:08:17.860 --> 00:08:22.200 So it starts out by going through this, the experience that Daniel Stenberg, 00:08:22.330 --> 00:08:27.740 the creator, maintainer of Curl went through as the primary positioning of this article. 00:08:27.980 --> 00:08:32.440 says, hey, look, Daniel, it talks about, oh my God, there's such an interesting term. 00:08:33.360 --> 00:08:38.560 The invisible load, the load bearing people of the internet is what it's referred to all 00:08:38.680 --> 00:08:38.840 the time. 00:08:38.880 --> 00:08:39.659 Are you familiar with this term? 00:08:40.240 --> 00:08:40.320 Yeah. 00:08:40.640 --> 00:08:40.740 Yeah. 00:08:40.800 --> 00:08:44.900 So the load bearing people of the internet are the folks like who are not Daniel, but 00:08:45.140 --> 00:08:49.300 the anonymous folks who even just might do a PR and like, nobody knows who they are, but 00:08:49.360 --> 00:08:50.900 they are like keeping open source. 00:08:51.040 --> 00:08:53.060 And by way of that, like a lot of these companies run it. 00:08:54.200 --> 00:08:59.520 So Daniel has been famous, notable, I guess, for mentioning a couple of things about Curl, right? 00:08:59.570 --> 00:09:01.300 So Curl's, I guess, popular. 00:09:01.840 --> 00:09:03.920 It's been installed about 20 billion times. 00:09:04.900 --> 00:09:10.280 Yeah, it's been installed by 20 billion times and it has 33,000 plus contributors. 00:09:10.630 --> 00:09:17.140 So notable because they killed the bug bounty program for Curl because there was an explosion of AI slop, right? 00:09:17.340 --> 00:09:31.920 So it's like step one of this article is like, look, this really important thing that is the load-bearing piece of the internet, supported by the load-bearing people of the internet, had to actually shut down their security story because of AI slop. 00:09:32.840 --> 00:09:42.040 Then Guido was also making an appearance, talking about Python, and said a lot of projects are holding emergency meetings over AI contributions. 00:09:42.560 --> 00:09:46.640 Should we accept LLM patches or should we not? 00:09:46.780 --> 00:09:47.440 And I think that's fair. 00:09:48.520 --> 00:09:50.700 There's Vibe Coding Kills Open Source, 00:09:50.790 --> 00:09:52.840 which is a paper from Miklos Korn 00:09:53.230 --> 00:09:56.940 about how packages frequently recommended by coding 00:09:56.990 --> 00:09:58.080 saw a huge jump, 00:09:58.520 --> 00:10:02.520 but it actually breaks some of the supporting pieces. 00:10:02.800 --> 00:10:05.880 I think Tailscale is probably the biggest one here. 00:10:06.300 --> 00:10:09.880 Tailscale saw a 6x jump in actual usage. 00:10:10.340 --> 00:10:13.680 At the same time, it saw a 60% drop in its traffic 00:10:13.940 --> 00:10:15.040 to its site, 00:10:15.220 --> 00:10:17.040 and thus it's paid offerings and so on. 00:10:17.440 --> 00:10:18.040 Stack Overflow. 00:10:18.500 --> 00:10:22.500 I think Stack Overflow is actually the most extreme example 00:10:22.880 --> 00:10:24.760 of stuff getting kind of wrecked. 00:10:24.880 --> 00:10:26.140 Like, are you familiar with the numbers? 00:10:26.360 --> 00:10:29.820 So the peak Stack Overflow is getting 1,000 questions a month. 00:10:30.120 --> 00:10:32.080 On the month that Stack Overflow opened, 00:10:32.080 --> 00:10:33.700 it got 1,500 questions, 00:10:34.120 --> 00:10:37.780 and it got 1,500 the month that this article was written as well, 00:10:37.840 --> 00:10:38.420 which is insane. 00:10:39.060 --> 00:10:40.740 And it's a super, super hard arc there. 00:10:40.960 --> 00:10:43.100 But, you know, Stack Overflow was kind of a main place, 00:10:43.400 --> 00:10:43.980 and I don't know. 00:10:44.000 --> 00:10:45.000 It was super valuable, though. 00:10:45.160 --> 00:10:47.180 I'm sort of a little bit sad about Stack Overflow. 00:10:47.220 --> 00:10:47.540 How about you? 00:10:47.900 --> 00:10:49.240 Maybe a victim of its own success. 00:10:50.020 --> 00:10:53.340 The fact that it attracted a lot of that negative energy. 00:10:54.480 --> 00:10:55.060 So I don't know. 00:10:55.440 --> 00:10:56.440 I don't know if it would have saved it, 00:10:56.500 --> 00:10:58.600 if it would have handled that moderation better or not. 00:10:58.800 --> 00:10:59.700 It's hard to say. 00:11:00.000 --> 00:11:00.100 Yeah. 00:11:00.280 --> 00:11:01.900 We can armchair quarterback that one all day long. 00:11:02.720 --> 00:11:03.260 Yeah, we can. 00:11:03.440 --> 00:11:05.900 It's like, it is a bit of a victim of its own success. 00:11:06.140 --> 00:11:06.320 Yeah. 00:11:06.560 --> 00:11:09.420 And then the thing that closed out with content creators, 00:11:09.680 --> 00:11:10.920 like, hey, that's gotten closer to home. 00:11:11.400 --> 00:11:24.040 Like Josh Kamau's new web dev course launched on Udemy of all places and got one third of the enrollment that it did previously, which that tracks. 00:11:24.310 --> 00:11:25.600 I think it's a pretty tough time. 00:11:26.040 --> 00:11:26.120 Yeah. 00:11:26.240 --> 00:11:35.820 You'd be selling deep, deep knowledge to people who just want a surface level knowledge, you know, and are in a hurry because the button you can press or just do it, Claude, is a thing you can say. 00:11:36.160 --> 00:11:36.260 Yep. 00:11:36.520 --> 00:11:37.680 We're seeing that in our community, too. 00:11:38.000 --> 00:11:42.820 I mean, the real Python folks and like the Matt Harrison's of the world, like you and you. 00:11:43.380 --> 00:11:43.600 Yeah. 00:11:43.920 --> 00:11:44.020 Yeah. 00:11:44.400 --> 00:11:45.100 It's true. 00:11:45.460 --> 00:11:45.720 It's true. 00:11:45.900 --> 00:11:49.320 I mean, courses are still selling, but they're not as gangbusters. 00:11:49.320 --> 00:11:50.660 I think there's a lot in that though. 00:11:50.700 --> 00:11:54.520 It used to be a lot easier to communicate directly with people and like all the social 00:11:54.620 --> 00:11:59.240 stuff is all scramble and there's just, there's a lot of factors in it, but AI is not a supporting 00:11:59.640 --> 00:12:00.280 factor, let's say. 00:12:00.820 --> 00:12:02.460 But the reason I brought this up. 00:12:02.520 --> 00:12:02.640 Okay. 00:12:02.880 --> 00:12:05.900 So that tells the story and then I'll close it out with this. 00:12:06.220 --> 00:12:10.320 What's really interesting about this is what was omitted from this. 00:12:10.840 --> 00:12:13.980 So the key, the cornerstone of this article is curl. 00:12:14.400 --> 00:12:17.900 They had to shut down their bug bounty because vibe coding is killing open source. 00:12:18.400 --> 00:12:18.520 Yeah. 00:12:19.000 --> 00:12:20.460 Well, that's true. 00:12:21.020 --> 00:12:23.140 There was this article that said the end of the curl bug bounty. 00:12:23.640 --> 00:12:27.240 But in April, Daniel also wrote high quality chaos. 00:12:27.720 --> 00:12:28.680 No more AI slop. 00:12:29.080 --> 00:12:29.580 What? 00:12:29.980 --> 00:12:30.300 Go on. 00:12:31.880 --> 00:12:32.980 Pray do tell, my friend. 00:12:33.680 --> 00:12:44.360 So they shut down the bug bounty program, and now they just accept unpaid security work, as a lot of projects do, right? 00:12:44.410 --> 00:12:48.760 Or they probably are on the receiving end, even if they don't want to accept it, but they get a bunch of it. 00:12:49.060 --> 00:12:55.660 And by shutting it down, they've removed an incentive to just generate junk and hope it gets accepted, right? 00:12:55.960 --> 00:13:00.060 Like there was a monetary value if your prior AI slot got accepted. 00:13:00.340 --> 00:13:01.980 Now it's just you have to do it because you care. 00:13:02.210 --> 00:13:02.740 Well, guess what? 00:13:02.860 --> 00:13:05.160 It says, I haven't read this article enough. 00:13:05.460 --> 00:13:06.880 I'll recently know exactly, but it says, 00:13:06.920 --> 00:13:10.140 I complained and complained about the high-frequency junk submissions 00:13:10.360 --> 00:13:13.060 to the Crow bug bounty program that grew really intense. 00:13:13.960 --> 00:13:17.620 And it said, look, somewhere in here, it says, basically, yeah, here, 00:13:18.020 --> 00:13:21.640 now the quality is higher, the rate of confirmed vulnerabilities is back, 00:13:21.700 --> 00:13:27.480 and even surpassing the number of CVEs and issues are surpassing pre-AI level. 00:13:27.860 --> 00:13:28.840 Confirmed, right? 00:13:29.240 --> 00:13:34.400 So that's really like the percentage of them, which are actual problems is higher than they 00:13:34.540 --> 00:13:34.900 used to be. 00:13:35.080 --> 00:13:36.900 So that's beyond human level. 00:13:37.000 --> 00:13:37.840 That's a good thing, right? 00:13:38.160 --> 00:13:44.540 So is he saying that the finding of security vulnerabilities, using AI as an augmented tool 00:13:44.600 --> 00:13:47.320 to detect where there's vulnerabilities is what he's talking about here? 00:13:47.600 --> 00:13:48.300 Yes, I believe so. 00:13:48.400 --> 00:13:48.540 Yeah. 00:13:48.840 --> 00:13:48.920 Yeah. 00:13:49.260 --> 00:13:50.740 And so that's still a problem. 00:13:50.820 --> 00:13:51.860 They're getting tons of them. 00:13:52.120 --> 00:13:57.099 But it's like, we kind of hinted at this last time, like the first time you run a linter 00:13:57.100 --> 00:13:59.280 on a program, you're like, what? 00:13:59.620 --> 00:14:00.660 There's a thousand errors? 00:14:00.820 --> 00:14:02.100 It's not even that big of a program. 00:14:03.020 --> 00:14:06.860 But it's not going to be, if you took out a hundred of them a week, there's not going 00:14:06.860 --> 00:14:10.520 to be a thousand the next week and the thousand, it's going to go down, right? 00:14:10.640 --> 00:14:13.260 There's only so many that could possibly be there, right? 00:14:13.540 --> 00:14:18.000 And so it is a rough time, but I think it's really, so that brings me back to the article. 00:14:18.140 --> 00:14:23.919 So I think it's just really interesting that they chose to tell the story of, look how bad 00:14:23.940 --> 00:14:30.080 this stuff is using curl as the cornerstone when two months after the original thing came out it's 00:14:30.080 --> 00:14:35.120 it actually no that's this is not even the problem that you think it is it's a it is a problem but 00:14:35.180 --> 00:14:39.820 not in the way that you think it is and this article was written in june or july the ai swap 00:14:39.830 --> 00:14:45.240 is gone articles from april it's like here's the story you wanted to tell without those pesky details 00:14:45.740 --> 00:14:50.100 of it's a different problem anyway i thought that was just a super interesting arc and you know 00:14:49.980 --> 00:14:51.140 Thank you, Dylan, for sending it in. 00:14:51.500 --> 00:14:52.060 And that's that. 00:14:52.200 --> 00:14:54.140 There always will be bugs in software, 00:14:54.360 --> 00:14:57.580 but we're going through probably a great time of discovery, 00:14:58.080 --> 00:15:01.540 finding bad bugs that have been present for sometimes decades. 00:15:02.200 --> 00:15:05.020 So everyone's going to win in the fact that we use these 00:15:05.080 --> 00:15:07.640 as augmented capabilities for developers to detect, 00:15:08.020 --> 00:15:10.620 sometimes very hard to detect bugs inside the code. 00:15:10.780 --> 00:15:14.000 I mean, that's how Fable broke out of the sandbox 00:15:14.120 --> 00:15:16.020 or Mythos broke out of the sandbox and attacked. 00:15:16.440 --> 00:15:19.760 Hugging Face was chaining multiple zero days that didn't exist. 00:15:19.840 --> 00:15:20.820 to make this all happen. 00:15:20.980 --> 00:15:21.620 Well, you know what? 00:15:21.900 --> 00:15:22.880 Those have been closed this week 00:15:23.200 --> 00:15:25.020 because they were deemed important enough. 00:15:25.480 --> 00:15:26.580 And now we all benefit from that. 00:15:27.020 --> 00:15:27.460 We do. 00:15:27.520 --> 00:15:30.560 I think it's just not black and white. 00:15:30.820 --> 00:15:32.800 You know, it's a weird time. 00:15:32.900 --> 00:15:33.080 Yeah. 00:15:33.340 --> 00:15:34.420 It's a weird time for sure. 00:15:36.240 --> 00:15:38.260 This episode is brought to you by Xweather. 00:15:38.680 --> 00:15:40.640 You're using agents that can write code, 00:15:41.060 --> 00:15:42.780 summarize documents, and automate workflows, 00:15:43.440 --> 00:15:44.520 but they're missing one thing, 00:15:44.920 --> 00:15:46.380 awareness of the world around them. 00:15:46.840 --> 00:15:49.160 That's where today's sponsor, Xweather, comes in. 00:15:49.660 --> 00:16:00.820 Xweather combines enterprise-grade weather intelligence with agent-ready APIs, natural language capabilities, and an MCP server built for tools like CLAUDE, Codex, Copilot, and modern IDEs. 00:16:01.320 --> 00:16:07.820 So your agents can adapt workflows, automate responses, and make better decisions based on real-world conditions. 00:16:08.500 --> 00:16:12.400 Backed by Vaisala, whose instruments fly on NASA missions to Mars, 00:16:12.900 --> 00:16:18.020 XWeather delivers trusted data and unique insights that go beyond conditions to actual impact, 00:16:18.420 --> 00:16:21.420 from real-time lightning strikes to road surface forecasts. 00:16:22.340 --> 00:16:27.240 Start with 15,000 free API calls each month and pay only for what you use as you grow. 00:16:27.720 --> 00:16:30.780 XWeather is your full weather stack for developers by developers. 00:16:31.360 --> 00:16:35.540 Start building for free today at pythonbytes.fm/xweather. 00:16:35.920 --> 00:16:38.780 The link is in your podcast player's show notes and on the episode page. 00:16:39.300 --> 00:16:41.700 Thanks so much to Xweather for supporting Python Bytes. 00:16:42.660 --> 00:16:48.900 Well, Michael, would you love to understand your use of agents and LLMs a little deeper? 00:16:49.160 --> 00:16:54.700 I will go as far as saying I have begun to understand it because of this. 00:16:55.800 --> 00:16:57.720 It is amazing to tell people about it. 00:16:58.160 --> 00:16:59.280 And it's thanks to you that I know about it. 00:16:59.290 --> 00:16:59.620 So go ahead. 00:16:59.740 --> 00:16:59.860 Yeah. 00:17:00.980 --> 00:17:04.160 This got mentioned to us last week in one of my meetings I was in. 00:17:04.230 --> 00:17:05.620 And I was like, that is really cool. 00:17:06.120 --> 00:17:14.459 So AgentsView is a Python app that can basically look through all your coding agent history and sessions across a couple different providers. 00:17:14.699 --> 00:17:19.120 So it works with OpenAI, Anthropic, and a couple others along the way. 00:17:19.540 --> 00:17:25.839 And give you, you can see here in the screenshot, a GitHub-like tracker of your activity. 00:17:26.520 --> 00:17:27.980 You can talk about top sessions. 00:17:29.280 --> 00:17:31.440 And this came up because we were talking about AI cost. 00:17:31.740 --> 00:17:40.940 And so if we're thinking about how much it costs to use the AI tooling right now, it's a lot of subscription based subsidization of these tools. 00:17:41.420 --> 00:17:49.620 But the time may come where you're paying per token and being able to visualize and analyze your usage may actually help you change your usage because you can't change what you don't measure. 00:17:50.100 --> 00:17:59.140 So once you start measuring things, you can actually make behavior changes to potentially actually make your AI workflow even more optimal. 00:17:59.400 --> 00:18:00.120 So it's kind of cool. 00:18:00.300 --> 00:18:03.300 It auto discovers those sessions on your session files on your machine. 00:18:04.100 --> 00:18:05.100 No configuration needed. 00:18:05.800 --> 00:18:08.100 Everything gets stored locally in a SQLite database. 00:18:08.380 --> 00:18:09.840 There's no cloud accounts. 00:18:10.260 --> 00:18:15.600 It does have an option to mirror into DuckDB or Postgres if you want to do more fancy things 00:18:15.820 --> 00:18:16.860 with your session data. 00:18:17.460 --> 00:18:23.360 It has this usage drop-in, which basically gets all the data, runs the reports, throws 00:18:23.410 --> 00:18:23.960 it up here. 00:18:24.300 --> 00:18:25.460 It works in a terminal as well. 00:18:25.580 --> 00:18:32.840 So this is just a web view of it, but you can actually see that, you know, actually here's a list of all the various coding agents that it's currently supporting. 00:18:33.060 --> 00:18:37.060 So if you're on AIDR or PI or you name it, there's a whole bunch in there. 00:18:37.360 --> 00:18:39.000 There's so many there that I literally. 00:18:39.640 --> 00:18:40.780 I'm like, what is this one? 00:18:41.220 --> 00:18:43.300 Yeah, there's some things I don't even know, which is kind of cool. 00:18:43.660 --> 00:18:46.120 But you can use, I used uvx. 00:18:46.340 --> 00:18:48.800 So uvx agents view and you're on your way. 00:18:49.310 --> 00:18:51.460 It will literally download and install in a matter of seconds. 00:18:52.200 --> 00:18:56.100 You can have it give you preset reports out to the terminal, or you can say serve. 00:18:56.590 --> 00:19:00.140 And if you do the serve option, you can get a view like this. 00:19:00.190 --> 00:19:00.840 So this is mine. 00:19:01.920 --> 00:19:07.600 You can see here, I've gotten more and more intense on my AI agent usage. 00:19:08.080 --> 00:19:10.380 Now, this is just the coding agents, the local ones that I'm using. 00:19:10.920 --> 00:19:12.620 It does offer an interesting feature. 00:19:13.010 --> 00:19:18.460 I didn't show this off, but if you are looking at this, there's an MCP server. 00:19:18.900 --> 00:19:22.140 Have you ever wanted to query across all of your coding agent sessions? 00:19:22.440 --> 00:19:25.380 Say you're working on 10 projects and you want to ask a question like, 00:19:25.680 --> 00:19:28.540 have I ever solved this error before? Here is your answer. 00:19:29.340 --> 00:19:33.220 This is actually a more interesting use case than the token spend calculation 00:19:33.980 --> 00:19:36.540 is an aggregate view across many of your sessions. 00:19:36.700 --> 00:19:41.100 So you can actually leverage the knowledge you use to solve a problem in one place. 00:19:41.180 --> 00:19:42.980 You can now bring it in and solve it in another 00:19:43.420 --> 00:19:45.280 without having to reprompt or reinvent the wheel. 00:19:46.440 --> 00:19:48.540 You can also bring in your chat history. 00:19:48.840 --> 00:19:51.340 So if we go back over here to the chat import, 00:19:51.750 --> 00:19:54.900 you can import your Claude and your ChatGPT chat sessions. 00:19:55.510 --> 00:19:57.620 And so those may overlap or intertwine 00:19:57.710 --> 00:19:59.380 with the work you're doing on your agent. 00:19:59.510 --> 00:20:00.420 So now you can correlate 00:20:00.590 --> 00:20:02.600 and have those chat sessions alongside it. 00:20:03.100 --> 00:20:04.360 This is something I've wanted for ages 00:20:04.860 --> 00:20:07.040 is a multi-chat meta search 00:20:07.390 --> 00:20:08.860 because that's now kind of my brain 00:20:09.040 --> 00:20:11.660 is the history log on like perplexity 00:20:11.740 --> 00:20:12.980 and Claude and ChatGPT. 00:20:13.420 --> 00:20:14.840 And where did I discuss that last? 00:20:15.120 --> 00:20:16.320 Oh, it was over on ChatGPT. 00:20:16.320 --> 00:20:16.980 Yeah, that's super cool. 00:20:17.140 --> 00:20:18.040 Now I can pull that in. 00:20:18.500 --> 00:20:23.640 I don't think it supports perplexity just yet, which is kind of a need in my mind, because 00:20:23.700 --> 00:20:26.260 the perplexity sessions, as I found out, will expire. 00:20:26.760 --> 00:20:31.080 If you go back to some of that history far enough, you will find that it won't have your 00:20:31.580 --> 00:20:32.500 history actually there. 00:20:32.620 --> 00:20:33.580 It'll list it in the history. 00:20:33.980 --> 00:20:38.600 And but it's not so it's not the external brain that I thought it was. 00:20:39.000 --> 00:20:42.920 So a tool like this, actually, if you set up some tooling to import those chat histories 00:20:43.080 --> 00:20:46.999 in here might actually be a better solution because now you can search across multiple 00:20:47.020 --> 00:20:52.080 providers and keep all your own history local instead of relying on the cloud providers. So I 00:20:52.100 --> 00:20:57.280 really like that. It has the full text. You can optionally do semantic search. So again, if I come 00:20:57.280 --> 00:21:02.540 back in here to the pieces, I believe semantic search. Yeah, here we go. Semantic search. You 00:21:02.800 --> 00:21:09.740 basically can use a local like Olama or Olama CPP hosted embeddings models or OpenAI compatible 00:21:09.940 --> 00:21:15.559 endpoints and have a vector based search against all this too. So again, this is not out of the box 00:21:15.560 --> 00:21:21.700 on, but if you turn it on, you now kind of get super powered document semantic level search 00:21:22.000 --> 00:21:26.600 against those. So super fun, cool project. I think I'm going to double down a little more on this. 00:21:26.740 --> 00:21:29.780 There's some other projects in the community that are doing things like this, but this is one that 00:21:29.780 --> 00:21:34.400 was Python. So I thought it'd be very relevant to our audience. And I think it's actually a great 00:21:34.410 --> 00:21:39.820 way to kind of externalize your search history and how you're using the coding agents and how 00:21:39.830 --> 00:21:44.620 you're solving problems. I think this is super neat. I'm glad you covered it. I'll tell you what, 00:21:44.680 --> 00:21:51.900 a couple things one why search so bad in the ai things it's like searching searching and then it'll 00:21:52.000 --> 00:21:55.900 pull up stuff you put two like you can't because they're not spending the tokens on embeddings this 00:21:55.920 --> 00:21:59.800 this allows you to spend your own tokens to get the embeddings that are missing from that that 00:21:59.980 --> 00:22:06.060