sebra

I've vibed my own orchestrator. It's a nextjs UI launching headless claude / codex sessions and reacting to changes in gitlab and linear. It's tailored to my previous manual way of working. Auto-adressess CI pipe fails, rebases dev, reviews, etc. Also contains an explainer / code review UI where I can ask follow up questions. A typical ticket flow goes through a bunch of agents with different prompts to ensure the end result is something I'm happy with. The goal is as little manual intervention as possible.

My approach is to build the builder rather than the feature. So if something is off in the output I try to adress it through the orchestrator. Might lead to a bit overfitting but I feel like the results are quite good too far. Also really nice to be able to remove useless stuff with a single prompt.

Building a custom orchestrator for my workflow was quite easy. Like 1-2 weeks of prompting and now my work mainly consists of refining linear tickets in a claude / codex cli and then reviewing the completed MR on the other side of my orchestrator (through the custom GUI).

I think the algorithm doesn't care about violence but for sex education I'm not as sure. It would make sense for social media to err on the side of caution against pornographic content, leading to censorship of sex education as well. Which is very unfortunate of course. But it's all about the advertisers and credit card companies in the end.

At least Anthropic have a max subscription for corporate. Codex is only pay-as-you-go pricing beyond the base plan. Hence I'm stuck with Opus for work for the foreseeable future.

The tweet misses the conclusion from the paper that handcrafted AGENTS.md might help. To me its no surprise that 100% vibed AGENTS.md are unproductive. Not reviewing your design docs is probably even worse than not reviewing your code? I've seen some AI-generated agents.md which were just plain wrong. No surprise agents perform worse after reading those.

I use AGENTS.md to make sure my agents loop effectively (tests, quality, etc). Not to describe the code / architecture.

It's because that price point is for individuals not for companies. So my company can't pay for the $100 plan unlike with Claude. Only pay-as-you go pricing is available for companies beyond the $29 plan which runs out for me in 2/5 hours. And pay-as-you-go is insanely expensive.

I'm working on a large (at least 300k+ loc) Django code base right now and we have 32 direct dependencies. Mostly stuff like lxml, pillow and pandas. It's very easy to use all the nice Django libs out there but you don't have to.

Thanks Dylan. Turns out my colleague actually had a teams call with you yesterday and the issue was confirmed and prioritised on your end.

You have a great product. Looking forward to get back to using it!

I've tried Greptile and it's pretty much pure noise. I ran it for 3 PRs and then gave up. Here are three examples of things it wasted my time on in those 3 PRs:

* Suggested to silence exception instead of crash and burn for "style" (the potential exception was handled earlier in code but it did not manage to catch that context). When I commented that silencing the exception could lead to uncaught bugs it replies "You're absolutely right, remove the try-catch" which I of course never added * Us using python 3.14 is a logic error as "python 3.14 does not exist yet" * "Review the async/await patterns Heavy use of async in model validation might indicate these should be application services instead." whatever this vague sentence means. Not sure if it is suggesting us changing the design pattern used in our entire code base.

Also the "confidence" score added to each PR being 4/5 or something due to these irrelevant comments was a really annoying feature IMO. In general AI tools giving a rating when they're wrong feels like a big productivity loss as then the human reviewer will see that number and think something is wrong with the PR.

--

Before this we were running Coderabbit which worked really well and caught a lot of bugs / implementation gotchas. It also had "learnings" which it referenced frequently so it seems like it actually did not repeat commenting on intentional things in our code base. With Coderabbit I found myself wanting to read the low confidence comments as well since they were often useful (so too quiet instead of too noisy). Unfortunately our entire Coderabbit integration just stopped working one day and since then we've been in a long back and forth with their support.

--

I'm not sure what the secret sauce is but it feels like Greptile was GPT 3.5-tier and Coderabbit was Sonnet 4.5-tier.

"The browser" -> "A browser" and this is not an issue

I have the same issue. I think it's because I'm adblocking because if I try in chrome with no adblocker it loads the ads instantly.

But eh either 5s of black screen or 60s of ads. I tried watching a 15 min yt video without adblock and it had 5 ad breaks with some unskippable ads.

That used to be semi-common for smaller transactions in Sweden but was made illegal. Not sure why, probably to fight tax avoidance.

At this point the cost of handling cash is way higher than handling cards and as no one in Sweden ever uses cash its no longer relevant at all anyway. Now many (maybe even most?) dont accept cash to avoid the cost of handling cash instead.

There are numerous things still missing in terms of async support. Most notably for me is DB transaction support which leads to most non-safe endpoints running on the shared sync_to_async thread and me having to separate my code into one async function calling another sync function wrapped in sync_to_async.

In fact if you look at the source there is a lot of async methods in the framework itself which just straight up calls sync_to_async e.g. caching. This doesn't matter as much as hopefully it will get added proper async eventually. But I think believing your requests wont block just because you're using async is a bit naive at this point in Django and the async implementation has been going for years.

Not to mention that the majority of third party libraries lack support for async so you'll probably have to write your own wrappers for e.g. middleware.