jmilinovich

This feedback is true gold. Thank you for this, I’ll make these tweaks to increase readability.

I also love just asking Claude to look at the repo and decide how it should use it. Have had a few times where it said “project not a good fit” which in and of itself is a good sign

Autoresearch is awesome for stuff that has really clear loss functions but most problems don’t have that. So if you’re trying to improve product quality or write great docs you can use goal.md

Appreciate that! This fast feedback loop is what I like most about personal software.

I had 30 broken Playwright tests and no way to tell which ones actually mattered. The problem wasn’t “fix the tests” — it was that there’s no coverage tool for test infrastructure trustworthiness. I had to build the ruler before I could measure anything.

So I wrote a file that defined a composite metric (four weighted components → one score), an improvement loop, and constraints. Pointed Claude at it. Went to bed. Woke up to 12 commits, 47 → 83.

The file became GOAL.md. The insight that surprised me: most software doesn’t have a natural scalar metric like val_bpb. You have to construct it. Documentation quality, API trustworthiness, test infrastructure confidence — these things have no pytest –cov equivalent. But once you build the ruler, the autoresearch loop works on them too.

The part I’m most uncertain about: the “dual score” pattern. When the agent is building its own measuring tools, it can game the metric by weakening the instrument. So the docs-quality example has two scores — one for the docs, one for the linter itself. The agent has to improve the telescope before it can use it. I think this is load-bearing but I’d love to hear if others have found different solutions to the same problem.

Easiest way to try it: paste this into Claude Code, Cursor, or any coding agent and point it at one of your repos:

Read github.com/jmilinovich/goal-md — read the template and examples. Then write me a GOAL.md for this repo and start working on it.

Happy to hear what breaks. The scoring script is bash + jq so it’s not exactly production-grade, and the examples are biased toward the kinds of projects I work on. More examples from different domains would make the pattern sharper.

Hi HN, My name's John and I'm a co-founder of Aesthetic. Our team spent 2 years running a design agency to understand what parts of the design process could (and should) be automated, and realized that designers spend a lot of time copy/pasting content into design templates. We thought their had to be a better way to do this with technology, so we built the V1 of Aesthetic, which is free for anyone to use as of today.

When a user enters a link to our platform, we're crawling the page for all of its meta data, and then using any Schema.org, Open Graph or Title tag markup as props that we interpolate into a series of React components. While this version of the app is limited to just working with web-based content via our app, we're going to be launch our API and enable users to plug in any dynamic content source as well as import their own templates for use.

We're excited to get this into the hands of the HN community for feedback and in hopes this can be helpful in automating people's marketing workflows today.

Thanks! John Milinovich john(at)aesthetic.com

The battle between the open and closed web continues. Interesting to see the pendulum swing from google pushing amp stories to aggregating the dark web. Wrote more about whether social killed the open web here: https://blog.aesthetic.com/blog/is-web-dead/

Pinterest <3's the longtail :) I'd imagine they have some good diagrams of trailer suspension designs, but almost certainly not as good as a solid YouTube video or blog post..

Pinterest is definitely an expert level content farm. For Google Image Search in particular, you're right that it does create a bit of a broken end user experience that feels a bit misleading.

It's surprising that Google hasn't cracked down on this yet or dealt with the obvious dupes problem (showing image result from Pinterest that came from website a vs showing result from website a).

haha, agreed... my post was syndicated with a Medium collection, and for some reason my post inherited its 'interesting' style..

CEO of URX here- thanks for your excitement Steve! We're definitely big fans of the app you built!

Currently we list out the availability of apps in our index by company and not by operating system; we’ll look into providing more detailed information.

Regarding IMDb: while the site index may be accessible and their app does have deep links, IMDb has not publicly exposed their deep link structure via meta tags. Currently, we're only indexing sites with these tags, but that will change shortly.

Deeplinks are hyperlinks to specific states within mobile apps.

For more background: - http://en.wikipedia.org/wiki/Mobile_deep_linking - http://blog.urx.com/urx-blog/deeplinking-in-3-steps

yes, that is WOI. Unfortunately, web standards haven't allowed for complete pop-up suppression cross-platform and cross-browser. Fortunately, there are clever ways to make it suck less.

yes, check out how the routing works: http://urx.com/getting-started

what would you be interested in doing with that data/index?

Agreed, and do think that will exist in the near-term as well. It's so exciting to see all the movement in this space.. it's a huge win in the name of user experience and I think that things will continue to move quickly here.

To be more specific, I was talking about "Facebook App Link index" which seems to be the way for a developer to find a deeplink based on a given weblink while going through Facebook.

Look forward to discussing further and seeing how we can work together here to connect the webs.

Similar to the proposed methods of Google and Twitter, Applinks is Facebook's proposal for how developers expose their deeplinks to the web via meta tag markup.

In addition to their meta tags, Facebook also released an API that lets developers send Facebook a request with a URL to receive a response for the equivalent deeplink. This solution works best for deeper native integrations and those who want to go through Facebook to get this mapping. Fortunately, URX omnilinks already support Applink tags and don't require any code-level integration for developers to start linking to other apps.

Basically, the tags are open source (good), much like URX's working proposal [1], and the implementation of using the tags goes through Facebook. Facebook requires a native integration and does get around browser flashes, but is much more difficult to integrate than omnilinks, where you simply put urx.io/ in front of any weblink.

[1] http://urx.com/documentation

I see your point. Will discuss with the team in earnest.

pretty sure the cause and effect was pretty clear there.. thanks for the feedback, though - will launch an A/B test against grumpy HN posts to validate ;)

totally agree- across the board, it's exciting to see how Google's implementing deeplinking across their properties.

see also: https://developers.google.com/+/mobile/ios/share/deep-link

Looks interesting, excited to give it a try. Out of curiosity, how does the integration differ between Apptimize and Taplytics? From the documentation, the spec's seem to be the same and I'd be curious to learn more about the differences in features. Thanks!

The goal here is to build a data framework that decouples dev time from an app marketer's need to try several approaches at once.

Would love to hear from HN any other strategies/suggestions that work in this vain.

Believe it or not, the owner at the time had a public WHOIS record. I sent them an email expressing interest in buying the domain, and they replied within 10 minutes. From there, we got on the phone and the rest was history.

yes, that is correct.

Apps don't use cookies but rather Apple IDFA and Android ID. This is one of the biggest differences between the app world and the web world.

We've built a free link generator (http://www.urx.com/about/urx_links) that helps expose schemas for apps that we know about. Please contact us (team@) with any that we're missing.

http://schemes.zwapp.com/ also has a ton of schemas, but is a bit unorganized and not completely reliable.

Deeplinking has been supported on Android since the beginning via Android Intents. Our ads work for both Android and iOS, and we are currently finishing our Android Turnpike library.

1- Definitely. this is something we can easily accommodate, depending on the advertiser's preferences (ex: mobile only vs. mobile second companies)

2- We've built our own link creator (see footer link on site) to make it easy to learn about other apps' url schemes and create links into their apps. these links can be used from anywhere, including the web. We're adding apps consistently, so let us know if there are any that aren't already there.

URX's ads are only shown to users who already have a given app installed. If a user uninstalls an app and sees a deeplink ad before our systems refresh, they will be directed to the app store on click.

completely agree- until then, URX's Turnpike Open Source library will have to do!