Can relate. The UK electronic eVisa app was pure garbage. The major redeeming feature of the UK civil service and the various regulatory quagmires is that they're effectively open source. You (or Claude) can read through their practice manuals or policies and find a work-around. But my goodness is it annoying until you figure that out. Another fascinating bit is you may think the various departments are connected but they are not. The nice looking UK Government Digital Service (GDS) Design System gives everything a veneer of connected competence, but under the bonnet, that slick UI signal is as reliable as a posh accent. Don't become a migrant if you don't have to.
OTOH, a lot of people who think they need a VPN really just need tunneling and authenticated access, so I can see the pitch for why Teleport's offering is a fit for many users who would otherwise consider tailscale.
Theres more to it. This falls into the realm of privileged access management. I think if you are critical infrastructure, financial institution, healthcare tech. This is non negotiable and it is part of your compliances. Just VPN do not cut it out. At adaptive [1], we do the same for server, databases and kubernetes clusters. It is a double digit billion dollar TAM.
Meredith's talk was extremely scripted, not very original and then she ducked out of taking any audience questions. Udbhav awkwardly stood there but seemed like he could have had much more to say. It was hard to watch.
Mona Wang's talk early on Day 2 wasn't recorded but was the polar opposite -- Original, off-the-cuff, engaging, and just fun to witness.
I find Jesse's style to be fascinatingly apt storytelling, and not until recently after reading many of his prior posts did I realise that he's not in fact some old-timer who lived through the history he often writes about. Bravo for putting this original content out there, JB. What's next?
Someone running cloudflared accidentally advertising a critical route into their Warp namespace and somehow disrupting routes for internal Cloudflare services doesn't seem too far fetched.
The irony is even deeper than it appears. According to current US copyright doctrine, if Claude genuinely did all the work with minimal human creative input, the Salt Bae dash of ASL2.0 is essentially decorative - you can't license rights that don't exist.
The research shows the US Copyright Office hasn't caught up with `claude` code: they claim that prompting alone doesn't create authorship, regardless of complexity. Without "substantial" human modification and creative control over the final expression, the code lands in public domain by default. Not that it matters here, but anyone could theoretically use Ronacher's library while ignoring the Apache 2 terms entirely.
What makes this fascinating is that Ronacher knows this ("Is that even valid when there's barely a human in the loop?") but published anyway. It's a perfect microcosm of our current predicament - we're all slapping licenses on potentially unenforceable code because the alternative is.. what exactly?
> What makes this fascinating is that Ronacher knows this ("Is that even valid when there's barely a human in the loop?") but published anyway.
That has very pragmatic reasons. People should be able to use this library, in my jurisdiction I cannot place things in the public domain. So there are two outcomes: the Apache 2 license is valid and you can use it, or it was impossible to copyright it in the first place and it's in the public domain. Either way you are free to use it.
I like the fact this mcp-debug tool can present a REPL and act as a mcp server itself.
We've been developing our MCP servers by first testing the principle with the "meat robot" approach - we tell the LLM (sometimes just through the stock web interface, no coding agent) what we're able to provide and just give it what it asks for - when we find a "tool" that works well we automate it.
This feels like it's an easier way of trying that process - we're finding it's very important to build an MCP interface that works with what LLMs "want" to do. Without impedance matching it can be difficult to get the overall outcome you want (I suspect this is worse if there's not much training data out there that resembles your problem).
reply