Over-engineered, verbose generated code

The code that I get back from the LLMs is confused - it pulls objects and methods from other languages, makes up APIs that don't exist, etc. It also doesn't understand the bigger picture of what all is possible in the system, so it gives me over-engineered code that could be done really simply using other tools.

codingdave · Nov 1, 2026 · comment in “Ask HN: Anyone here still writing code for reasons other than fun or learning?”

Haha I always make it a point these days do handwrite these things. I think everyone is tired of overly verbose perfectly written AI text.

pancomplex · Oct 26, 2026 · comment in “Show HN: Hob - Open-source, on device alternative to Claude Cowork and ChatGPT”

OTOH it's a godsend for setting up a local dev machine or something like that, where you need to run different vendor provided packages. but I'm tired of all the mindless over-engineering (which unfortunately the LLMs trained on).

MikeRichardson · Oct 25, 2026 · comment in “Docker is weird”

> steer Claude away from over-engineering parts of the system in response to hypothetical concerns. This is my biggest problem with opus 5.5.

xnorswap · Oct 16, 2026 · comment in “Redesigning Source Control at Anthropic”

In my experience, AI code is really hard to maintain and overly complex which creates unexpected behavior and hard to catch bugs. I inherited some vibe coded repos in my company and it's been really painful removing all the unnecessary complexity

gonzalohm · Oct 13, 2026 · comment in “Epic Games tells employees to use AI to get Disney game done”

Trying to optimize tools in the age of AI is like putting lipstick on a pig. - It's more verbose than most other languages; waste of tokens, context window, time and AI reasoning capacity.

jongjong · Oct 1, 2026 · comment in “How to speed up the Rust compiler in September 2026”

Like JSON, MCP solves a problem without being overengineered. Both are painful for the overengineering type.

postalrat · Sep 21, 2026 · comment in “MCP was always a bad idea?”

You can prompt them to use any tone , be as concise or verbose as you like. This doesn't work by, effectively, Anthropic's own admission.

zamalek · Sep 17, 2026 · comment in “I Don't Like LLMs”

Claude doesn't have access to I had Gemini do some "deep research". When I fed the verbose report back to Claude it basically said it was a bunch of "hallucinated bullshit".

josefresco · Sep 17, 2026 · comment in “I had Gemini train its own replacement for $9”

It's extremely verbose, hard to read, often repeats code instead of factoring it into reusable components. And yes, sometimes it's also buggy.

roblabla · Sep 15, 2026 · comment in “How much of F-Droid is LLM generated?”

Also, while this little demo is certainly exaggerating the issue, I do find working with Claude to sometimes get quite verbose and tiresome. I doubt I would struggle this much to get it to change a button color, but the patterns of speech, the endless lists, the over-explanations, and the whole song and dance of trying to get it to make the change you want without side-effects is frustratingly familiar to me.

cub-creature · Sep 9, 2026 · comment in “Claude, change the "Add to Cart" button to blue”

End with one concrete next step. I mean maybe I have ADHD idk i'm not diagnosed, I don't generally have issues understanding Claude's output (even though I do think it's unnecessarily verbose), but I think this one is a general good advice which Claude ignores for some reason, and is especially annoying during debugging sessions, where it asks for 3 outputs, I paste one for it to process while getting the others, an…

black3r · Sep 9, 2026 · comment in “I-have-ADHD: A skill to stop coding agents from burying the answer”

Sounds useful. I'm often struggling with how extremely verbose Claude is.

mcv · Sep 9, 2026 · comment in “I-have-ADHD: A skill to stop coding agents from burying the answer”

Its not good at not making mistakes, but what it produces is structurally quite nice, not over-engineered (looking at you Sol) and its personality isn’t annoying (looking at you Claude).

MrBuddyCasino · Sep 2, 2026 · comment in “Gemini 3.8 Flash and 3.8 Flash Cyber”

What am I missing here. Commenting because I am struggling with this too, claude code seems to be so verbose no matter how I prompt it.

smashed · Sep 2, 2026 · comment in “Claude Fable 5.1 and Claude Mythos 5.1”

Has anyone been able to steer Claude so that it doesn't use this kind of vocabulary and super verbose code comments? We use it at work and the language and verbosity has been one of the most annoying things.

stefandesu · Aug 28, 2026 · comment in “Show HN: The load-bearing vocabulary of Claude”

LLMs vastly overengineer code (I believe they're intentionally designed to do so). I really struggle to make LLMs generate lean code.

pizza234 · Aug 25, 2026 · comment in “The state of AI in 2026: On the road to ROI”

I'm largely with everyone that Claude is much too verbose and uses too much annoying flowery language.

Myrmornis · Aug 22, 2026 · comment in “Claudette: Make Claude stop talking like a BuzzFeed article”

The one caveat is opus has an absolutely awful voice. The way it writes is annoying and verbose.

preg_match · Aug 21, 2026 · comment in “Kagi added a setting for removing paywalled links from search results”

When Claude overengineers something, I catch it during the planning phase, and it implements plans faithfully. 5.6 Sol is simply unreliable.

enraged_camel · Aug 18, 2026 · comment in “GPT-5.6 Sol Pricing Cut by 50% on OpenRouter”

But when it comes to working in a corporate environment, I have to ultimately follow a Jira ticket or a Asana ticket. Now, when I'm following those tickets, I am able to do the work, but it's not always up to the mark because after all I found AI to hallucinate, Unnecessary code is being written, even though a function for that is already there and can be reused, no proper testing apart from unit test being done so …

berwinsingh · Aug 4, 2026 · story in “Show HN: OldHand A Claude/Codex plugin to verify the development flow end-to-end”

For typing in specific bash commands, it's fine, but agentic coding is much more verbose. Ideally we could use the phones native speech to text, but this doesn't work on a terminal UI - it needs a real, editable text field so it can insert and revise words as you speak, something a terminal UI can't provide.

mdp · Aug 3, 2026 · story in “Show HN: Driftty – mobile focused web TTY”

Just an half hour ago, TomasBM wrote in another thread [1] why people first want to filter out AI slop, which IMHO fits perfectly: > Getting those verbose, AI-authored walls of text is very annoying, especially when you're expected to thoroughly review it.

mschuster91 · Jul 1, 2026 · comment in “The first early human eggs from stem cells”

It's a fair policy. Getting those verbose, AI-authored walls of text is very annoying, especially when you're expected to thoroughly review it.

TomasBM · Jul 1, 2026 · comment in “Godot will no longer accept AI-authored code contributions”

Then: a nice but verbose literature overview of my chosen topic <blink>BUT it includes at least one hallucinated reference!</blink>

raphman · Jun 30, 2026 · comment in “Claude Science”

Sure, you could hide it behind a --verbose flag, but then if it fails half way through, it's too late to add the flag. Even a user who doesn't understand the output could then copy/paste it into Google or an LLM to figure out what happened.

wtetzner · Jun 18, 2026 · comment in “Lore – Open source version control system designed for scalability”

Now it's much more verbose. My biggest gripe is that it basically stopped listening to me, since "upgrading" to Gemini, which is frustrating because I've used it to control the Hue lights for the past decade.

themythfable · Jun 16, 2026 · comment in “I Fired Google”

I recently discovered that Gemini can manipulate the navigation app, and a prompt like "Mute alerts" works, which is kind of cool. However like OP said, it's incredibly verbose, which is super annoying.

elbasti · Jun 16, 2026 · comment in “I Fired Google”

> IMHO the overly-verbose default style of LLMs is the most annoying part of interacting with them, and I wish their masters would just tell them to be terse by default.

lelanthran · Jun 12, 2026 · comment in “AI agent bankrupted their operator while trying to scan DN42”

IMHO the overly-verbose default style of LLMs is the most annoying part of interacting with them, and I wish their masters would just tell them to be terse by default.

userbinator · Jun 12, 2026 · comment in “AI agent bankrupted their operator while trying to scan DN42”

I don't care if they're AI-made, as long as they're short. I'm a very verbose person, and if I don't make an effort at being concise, I'm just as annoying as the average AI.

sshine · Jun 11, 2026 · comment in “If you are asking for human attention, demonstrate human effort”

The same goes for AI-generated conversations; verbose responses are the default behavior, and models are incentivised to keep that output-token ratio. Too easy to catch/notice, pretty annoying.

pravj · May 21, 2026 · comment in “Throwing AI-generated walls of text into conversations”

I was getting a lot of overly verbose annoying emails at my one workplace and I swear they probably had started to use AI to generate a lot of it.

Fogest · May 21, 2026 · comment in “Throwing AI-generated walls of text into conversations”

I think that nowadays with vibe/agentic coding, high performance Python-like languages become ever more important. Directly using AI agents to code, say, C++, is painful as the verbose nature of the language often causes the context window to explode.

dllu · May 8, 2026 · comment in “Mojo 1.0 Beta”

What would be genuinely useful? What would be over-engineered, annoying, or harmful?

skeedle · Apr 30, 2026 · story in “What do you want to see in a next-generation GitHub in the age of AI?”

System prompt to make Claude less verbose reducing coding quality (4 days - better) All this to say... the experience of suspecting a model is getting worse while Anthropic publicly gaslights their user-base: "we never degrade model performance" is frustrating.

jryio · Apr 23, 2026 · comment in “An update on recent Claude Code quality reports”

Pointing it out because it may be an indicator it was rushed and over-designed, which would explain it's relative complexity compared to ~anyone else. * I hate talking about myself, but hate it less than being verbose and free-associating without some justification of relevant knowledge: quit Google in late 2022 to build a Flutter all-platform LLM client, based on llama.cpp / any 3rd party provider you can think of.

refulgentis · Apr 14, 2026 · comment in “The M×N problem of tool calling and open-source models”

Its so verbose... I wish AI could simplify things.

srslyTrying2hlp · Apr 14, 2026 · comment in “Schools Never Taught Critical Thinking: AI Exposed the Lie”

TBH I'm not sure this is where dev is going. It's pretty annoying to have to read what can be pretty verbose LLM output on a 5 inch screen.

giwook · Apr 6, 2026 · comment in “AI agents over SSH from your phone. Starting to think this is where dev is going”

The problem: I wanted to pipe PR review feedback into AI coding agents, but the GitHub REST API doesn't expose thread resolution status, and the raw data is verbose. Copy-pasting from the UI wastes tokens and loses structure.

Nittarab · Feb 25, 2026 · comment in “Gh-PR-reviews Fetch GitHub PR reviews as compact Markdown coding agents”

On the overengineering threshold: For me it became necessary after about 200 dispatches. One of my agents hallucinated a dependency, a second agent "fixed" the resulting test failures by creating the missing module, and by morning I had three rewritten files with clean commits built on something that should never have existed.

vincentvandeth · Feb 21, 2026 · comment in “Ask HN: What makes AI agent runtime logs defensible under adversarial audit?”

If the AI claims a file is "Tier 3" (Complex) but the complexity is < 15, we automatically downgrade it. (This prevents the "hallucination" problem where AI thinks a long file is complex when it's just verbose).

dhruv0206 · Feb 11, 2026 · story in “Show HN: DevProof – Verified developer portfolios using Code Complexity and AI”

They're powerful, but setting up the whole AST pipeline and that plugging system is for me an unnecessary complexity. On the other hand, you have things like @next/mdx which are cool but too page-focused and doesn't work on the client side.

aypineau · Jan 29, 2026 · story in “Show HN: A cross-framework Markdown/MDX parser to simplify content management”

I built the desktop app in ~14 days using heavy AI assistance. What worked surprisingly well was forcing a planning gate before any code generation: having the model produce a phased implementation plan first reduced over-engineering and large hallucinated files.

mckabue · Jan 4, 2026 · comment in “Show HN: Blober.io – A local-first cloud transfer engine built in 14 days”

It can take simple directives and turn them into hierarchical Markdown lists that have multiple bullet points. It's annoying and overly verbose for humans to read, but the repetition and structure seems to help the LLM.

Aurornis · Dec 22, 2025 · comment in “Scaling LLMs to Larger Codebases”

GPT-5 was the first model that occasionally produced code that I could push without any changes Claude still tends to add "fluff" around the solution and over-engineer, not that the code doesn't work, it's just that it's ugly

SamDc73 · Dec 18, 2025 · comment in “GPT-5.2-Codex”

• “How do I give my team access without accidentally giving them access to ruin my budget?” And then there’s AI Slop — unnecessary tokens, verbose responses, hallucinated filler text, or redundant reasoning chains that waste tokens without adding value.

mdzakki · Dec 12, 2025 · story in “Building an AI cost-optimizer and AI Slop Prevention tool Looking for feedback."”