frontpage.
newsnewestaskshowjobs

Open Source @Github

fp.

Open in hackernews

Ask HN: Does spec-driven development still pay off with frontier coding models?

3•sarangk90•1d ago
As coding models get better at working from a rough goal, I’m wondering how much value there is in writing a detailed spec up front.

If you use these models on real projects, when has a spec improved the outcome? When has it added overhead? Have you shifted toward shorter briefs, tests, or iterative feedback?

Comments

bbor•1d ago
I'd argue that you 100% need a spec --ideally tied to some high-level unit tests-- and should have a plan, but I don't think there's any strict need to write any of those three by hand; the critical step is where you review it and request changes if needed.

Doing it all manually better guarantees quality of course, so makes sense for a lot of problems. But if you're looking to move a bit faster with a bit more risk, I'd sacrifice the writing step without sacrificing the reading & validating step.

docenttx•1d ago
I tend to agree with bbor. I use the agent to build the spec, usually from unfiltered audio to text data dump. The spec also allows you to manage context by pre-defining scope. I would argue managing context is more important than a spec at a certain quality level of the spec. A high-quality spec should inherently help manage context. If you are leaning into full end to end automation, there is no substitute. I'm not sure "better" is the lens. A more capable model also has more paths of interpretation that it can conceivably take based on the framing of your rough goal.

Tell HN: Codex Is Down [fixed]

55•minimaxir•1h ago•41 comments

Ask HN: Who's still keeping a DOS machine up because the business depends on it?

62•mlaux•4h ago•46 comments

Ask HN: Hypothesis: Cellular providers are deprioritizing voice calls?

12•tkdb•1d ago•24 comments

Ask HN: Any nerds out there who've read a lot of research papers?

14•shriy-singh•1d ago•13 comments

Ask HN: Is Opus 5.5 another step change?

4•consumer451•1h ago•3 comments

Ask HN: IEEE expulsion of Fei-Yue Wang?

6•sathomasga•1d ago•1 comments

Tell HN: OpenAI $500 ProMax plan listed in API

19•GustavHartz•15h ago•22 comments

Ask HN: How do you deal with cyber attacks once you scale?

4•cehgmaster•1d ago•2 comments

FYI: OpenAI "spend limits" aren't always limits

2•iambateman•2h ago•0 comments

Tell HN: Chrome 154 corrupt Fontconfig cache and crash KDE Plasma

3•vdfs•3h ago•0 comments

Ask HN: What vendor you'd switch out if a better option was available?

3•jviotti•6h ago•9 comments

Ask HN: LM Studio Bionic with Python?

2•HoldOnAMinute•4h ago•0 comments

Ask HN: Which model do you use for work?

3•george_max•5h ago•3 comments

Ask HN: Has anyone built a leaderless multi-agent system?

2•har-ki•6h ago•0 comments

Ask HN: Anyone else's OpenAI and Claude API keys mysteriously leaked last night?

2•par•9h ago•0 comments

Ask HN: How do you use LLMs to secure your code and services?

3•cpa•12h ago•0 comments

Ask HN: Migrating personal email domain from Gmail to iCloud?

4•osener•16h ago•4 comments

Ask HN: Where to look for jobs?

4•0x737368•22h ago•1 comments

Ask HN: Is it impossible to disable Siri on macOS 27?

153•semidror•4d ago•87 comments

Ask HN: Are .shop domains down?

5•esher•1d ago•7 comments

Ask HN: How do you feel about the new $500 OpenAI subscription?

2•preommr•1d ago•8 comments

Ask HN: Does a startup need both founder and company X accounts?

3•OnionLayers•1d ago•0 comments

Ask HN: Do you cuss at your agent?

3•ak39•1d ago•8 comments

Ask HN: Folks familiar with infections after TKA

3•zafka•1d ago•0 comments

Ask HN: Does spec-driven development still pay off with frontier coding models?

3•sarangk90•1d ago•2 comments

Ask HN: What job boards are good these days?

21•phendrenad2•2d ago•16 comments

Tell HN: Substack obfuscating text to break reading mode

16•xnx•3d ago•4 comments

Anyone using a Wispr Flow alternative that is non-cloud?

21•itsjeremiahs•3d ago•17 comments

Tell HN: PSA: AI can make old games run on 64-bit macOS

5•Razengan•13h ago•0 comments

When Claude CLI asks for feedback, responding authorizes conversation capture

25•mmaunder•4d ago•3 comments