Rendered at 08:40:47 GMT+0000 (Coordinated Universal Time) with Cloudflare Workers.
OptionOfT 13 hours ago [-]
A bunch of these should be enforce with linting, that way people who still hand-craft code get the same kind of feedback, e.g. Always use {}, even on a one-line "if" statement. & Keep function names short. Less than 30 characters.
Then this one really is a pattern that creates a lot of churn:
- Add a small, to the point, comment to explain what the block does and why. Use examples when possible. Propose ASCII drawings to explain complete systems.
The what _is_ the code.
hawk_ 10 hours ago [-]
I forbid my agents from adding any comments. I review the code and add comments manually. If I can't understand something despite having the context then I throw away the code instead of having an LLM generate comments to explain what it did. This way the code stays readable/debuggable by humans.
solatic 2 hours ago [-]
> This way the code stays readable/debuggable by humans.
Please take the following as expressed with genuine curiosity: Do you not use an editor with syntax highlighting and collapsible comments?
At least on JetBrains you can configure the editor to collapse all comments on open and to have the comments displayed in a low-contrast color. This way, LLMs add a bunch of comments, but it doesn't affect your actual experience in trying to read the code. If you encounter code that seems inexplicable, then and only then would you expand the comment to see if that helps you understand.
metek 1 hours ago [-]
LLM comments for code are almost unfailingly completely redundant or impenetrably verbose bordering on word salad.
shunia_huang 36 minutes ago [-]
Sometimes I try to add comments in a new session and the agent just don't have enough context for it to give a comprehensive sentence with full context on the why, then the agent will just describe what it does.
Human comment is in another level to answer the questions mainly like "why do it like this" for the later collaborators or the forget-ed self, so the important blocks live when it is needed and can be eliminated when it does not.
jodleif 18 minutes ago [-]
Also, the language model might not fully understand the code then add a comment, then the next iteration will treat assumptions in the comment as the truth.
digitalPhonix 34 minutes ago [-]
I think “This way the code stays readable/debuggable by humans” is a proof by example (not that the generated comments are necessarily bad).
If the human can read/understand it well enough to comment it, then it is readable by humans.
watwut 26 minutes ago [-]
Misleading and hard to read comments are worst then none at all for readability. If he did what you suggest, he would end up with tons of bad javadoc.
imagetic 3 hours ago [-]
Mad props to you for that. Smart.
robby_w_g 9 hours ago [-]
How do you stop LLMs from making comments? In my experience, LLMs treat requirements for code output as suggestions
embedding-shape 2 hours ago [-]
Add "Don't add any code comments anywhere" to your system prompt.
If the model doesn't follow this, you want to start using a better model ASAP, because SOTA models for the last year or so, been able to following this without an issue.
yehoshuapw 59 minutes ago [-]
I've come recently across arxiv 2604.20911
which claims "do" rules persist much better then "don't" rules.
embedding-shape 45 minutes ago [-]
Alright, what you'd put instead of "Don't add any code comments anywhere"?
I agree with the general guidance, but it's a general one and not applicable for everything. Some things cannot be expressed in a "do" way rather than "don't".
yehoshuapw 36 minutes ago [-]
I don't know. I fully agree, and never did really try out this in depth yet.
I suspect rules with negations are not the same as don't rules, but unknown if really true, if so:
"when writing code do not add comments, code should not need it" may work
embedding-shape 30 minutes ago [-]
Your proposal is still a "Prohibition-type constraint" that the paper you linked earlier say "isn't good".
Some of these constraints we want simply aren't possible without adding "do not" somewhere in the line, even if you prefix/suffix it with other stuff, as you noticed yourself :)
miki_oomiri 8 hours ago [-]
Ask the agent to write a script to run after each changes, against the newly added code.
Use that script as a super linter.
That’s the only way I found to strictly enforce some rules, like the no comments rule, without enforcing them against my own changes or old code.
eru 4 hours ago [-]
You could run the script mechanically against the diff (assuming you use version control). No need to rely on the agent.
ks2048 7 hours ago [-]
If you don’t trust the code to write a decent comment, why trust it write good code?
Of course, ensuring compilation or other checks can verify some code, which it can’t do for comments. But comments still serve the same purpose as human comments.
Infernal 5 hours ago [-]
If I understand correctly, it’s not that the LLM can’t write a good comment, it’s that you want to be able to interpret and understand the generated code without comments - and in that process end up writing comments yourself.
9 hours ago [-]
lkjdsklf 9 hours ago [-]
That seems like a really smart workflow
I wish my coworkers would adopt this.
I’m sick of reading a fucking Charles dickens novel for every fucking tiny function
jamesfinlayson 3 hours ago [-]
Ugh, this. Had a workmate recently churn out 4,000 lines of code using Claude and I'm sure half of it was just comments.
Mtinie 7 hours ago [-]
[flagged]
figmert 10 hours ago [-]
Right. I've really struggling to get AI to stop explaining the what. It seems to add it to the commits, PRs, code, wherever it feels like. I've put in multiple places to not write the "what", but the "why", and in multiple ways, but it still does it in one or other place.
jaggederest 10 hours ago [-]
The best way I've found to solve this is using LLM as CI - use a small cheap model to inspect the diff and look for those kinds of comments. Prompt left to the observer but using `claude -p` / `codex exec` gets you a lot cleaner output usually, and makes robots fight robots instead of you constantly having to reprompt and it ignoring you.
5thaccount 8 hours ago [-]
I've reached that point as well. Is there a preferred model and prompt you use for that?
jaggederest 2 hours ago [-]
[flagged]
getnormality 13 hours ago [-]
I would never tell an agent to write "what does the code do" comments. Their default comments are already way too fluffy.
saghm 10 hours ago [-]
But then you don't know about where the load bearing seams are!
JSR_FDED 5 hours ago [-]
Honestly? That’s the kicker.
telotortium 7 hours ago [-]
Hmm, I guess everyone here is using Claude? I find that Sol is much more restrained, to the point where I have a prompt to tell it to add short comments for things that are not obvious. Really, I find the verbosity problem to be worse in tests. I regularly prompt my review agent to remove tests used only for scaffolding to write the code in the first place. The agent is in a way following strict TDD, which reminds me why I don’t like TDD, even though some of the generated tests can be useful.
sampullman 4 hours ago [-]
Sol adds comments when I use it, but it's not nearly as verbose as Opus 5/Fable.
For the latter I'll often include an example of a comment it wrote, along with my own rephrasing, and tell it that "future readers will understand code context; good naming is the best documentation". This works alright if I include in the actual prompt, but annoyingly it often doesn't in CLAUDE.md or memory.
throwatdem12311 9 hours ago [-]
I tell the agent to NEVER write comments in the system prompt and it ignore it like 90% of the time. RLHF is a helluva drug.
embedding-shape 2 hours ago [-]
What models are you using? I've had "Never add any new code comments" in my system prompts for like a year at this point, seems every model above 14B picks this up just fine. Are you using llama 7b or similar for testing this?
bombcar 6 hours ago [-]
We trained agents on millions of pages of documentation telling them to write good comments and good code and then we tell them never to write any comments.
It’s almost the “we built a robot who loves to play Sonatas and gave it no hands” type of thing.
rustystump 11 hours ago [-]
I added to the memory, system prompts, and the prompt itself and every soa model still litters code with the most inane useless crap. I will then get code to review from a coworker using fable/opus. It has more lines of comments then code.
Maybe I am some god tier code reader (i am not) but i dont think i have ever found a comment in code to be useful in my day job. That isnt true, i once came across
// submit to the dark lord
Above the function that sent a payment to PayPal for processing. It made me laugh so I let it be.
danielheath 10 hours ago [-]
Most useful code comment I have encountered read:
“””
After you give up on trying to refactor this code, increment the following line accordingly. HOURS_WASTED_HERE=26
“””
12 hours ago [-]
culi 11 hours ago [-]
My biggest pet peeve with agents is when people beg their (non-deterministic) agents to do something that a lint rule could've accomplished
irishcoffee 10 hours ago [-]
Seems like 80% of agent use boils down to: grep | sed -i
Which is kind of cool if you’re unaware enough to know to do it yourself.
Oh, and find. Agents use find a lot.
eterm 10 hours ago [-]
So it turns out that a lot of these unix utilities have such bad UX that having a tool that knows how to really leverage them feels like a superpower.
If you've ever used an LLM to deal with ffmpeg you'll know exactly what I mean.
drfloyd51 8 hours ago [-]
It is incredibly difficult to make an UX that can beat simply typing what you need in your own words.
You think of what you need, and you type it. No need to even ask “what options should I use?”
skydhash 7 hours ago [-]
I would rather bet that people don’t know that their problem has been solved for ages. Either they don’t know about the tools or can’t make the leap to think of using something like awk or sed to quickly script out their use cases. Or even quickly draft up a quick function/plugin in something like vim, emacs, sublime,…
In “The Pragmatic Programmer”, the power of unix tools and editor fluency is well argued. There are plenty of other books like “Unix Power Tools”, “Small, Sharp Software Tools”,…
selcuka 9 hours ago [-]
Pi even installs ripgrep and fd if it can't find them in the path.
telotortium 7 hours ago [-]
It turns out that thinking about and executing these commands at a superhuman speed is, to ape Claude, the real unlock.
mpyne 5 hours ago [-]
Yeah, I've actually found in my own testing and usage of LLMs that this is where I get a lot of benefit. I already have fd, ripgrep, etc. installed and know how to use them, but it's not hard to tell the LLM to do it and it often finds things just as well. Or even better.
It's especially handy on modern style code where things get broken up across a multitude of files based on convention.
duttish 2 hours ago [-]
The cost of custom linters has like any other code dropped through the floor. I'm sprinkling all kinds of linters over my latest projects. It seems some people are still sleeping on this, expecting great code from the agents.
They're fast and deterministic and I run them in git pre-commit.
embedding-shape 2 hours ago [-]
> expecting great code from the agents
Yeah, I'm not sure what people are thinking. I keep reading stuff from folks like "I wish models had more common sense" and "I wish they wrote better code", not realizing this is 100% in your own hands, always been. There is no such thing as "clean code" that every programmer agrees on, you have your own subjective opinion and "good" taste about the code, instruct the models to follow it! And automate it while you're on it.
tyre 10 hours ago [-]
Mine says what I tell engineers:
> Write in-code comments that describe _why_ code or a class does what it does, but not _what_ it does. The "what" should be self-evident.
arialdomartini 10 hours ago [-]
Incidentally, I'm from the opposite school and consider every “if” followed by a braced block a smell.
If a conditional body needs a block, it's doing enough to deserve a name, so I promote it to a single named call, à la "Extract till you drop".
AdieuToLogic 8 hours ago [-]
Another phrase for this is "functional decomposition", which usually is a good thing.
Better yet is to identify conditional execution paths as early as possible in order to obviate conditionals in the call tree. For example, identifying a "create a new something" verses an "update an existing something" based on the workflow initially invoked greatly simplifies service and/or persistent store logic.
what 2 hours ago [-]
So you’ll make a two line function to replace a braced block? Seems kind of unhinged.
0xfeba 10 hours ago [-]
> Propose ASCII drawings to explain complete systems.
Agreed. One of the first rules I toss into Biome is `noNestedTernary` - LLMs seem to adore completely unreadable nested expressions.
throwatdem12311 9 hours ago [-]
Linters and static analysis -> setup as hooks in your harness. Don’t rely on CLAUDE.md because it’ll ignore it a lot.
> ASCII drawings in code
Please don’t this is super obnoxious. Make proper diagrams and kee them in knowledge base. Link out to them if you need to and let the agent fetch them via MCP or API or whatever if it wants them.
4 hours ago [-]
skydhash 8 hours ago [-]
> The what _is_ the code.
Even the why sometimes shouldn’t be a comment, unless it’s very immediate to the code itself. What’s often more necessary is a high level overview of the design of the solution, because that’s what drives the design of the code and link disparate section. Especially the glossary , which you let you understand the name of the symbols (variables, struct. functions,…) used in the code.
It’s like learning the culture associated to a foreign language instead of trying to translate each single word with a dictionary.
_boffin_ 10 hours ago [-]
One thing I don’t get with a lot of these agents.md and other skills are… why not throw as much mechanical checks and other stuff at the repo to constrain as you want instead of asking a non-deterministic agent (squishy or non-squishy) to maintain it.
With the mechanical routes, we get checks, failures, and so much more. A bit wild to me.
Make an agent operate within defined constraints and yell at it when it doesn’t.
IanCal 7 hours ago [-]
Do both. Instructions help avoid the first pass from making the same mistakes.
> Make an agent operate within defined constraints and yell at it when it doesn’t.
And tell it what the constraints are.
_boffin_ 3 hours ago [-]
[flagged]
andai 7 hours ago [-]
> - Keep function names short. Less than 30 characters.
Recently I asked GPT to port a browser game to Rust. It voluntered this gem:
I like long function names as long as they're still explanatory and not distracting.
swiftcoder 15 minutes ago [-]
This is an unfortunate side effect of bridging an API from a language that supports function overloading, into a language that does not
krackers 6 hours ago [-]
They laughed at objective-c back then, now the shoe's on the other foot
egorfine 14 minutes ago [-]
So much this. DoThisWithThatAndThatAndThis is a objective c convention and I love it so much that I use it everywhere.
Although a bit of taste is required to keep code legible and pleasant.
YuechenLi 12 hours ago [-]
Since we are sharing our AGENTS.md, I thought I'd share my own, because most of the time, this is pretty much all you need for LLMs to write good code, everything else can be added per project:
----
*Convergence rule*
Every substantial task must end in exactly one of three states:
A. Success
The intended capability works in the real path and the real motivating case materially improves.
B. Meaningful progression
The capability is not complete, but one genuine blocker is removed and the next blocker is isolated with evidence.
C. Honest stop
Further work would require overbroad scope expansion, excessive debt, brittle patching, or tangled logic. Stop and report the reason with concrete evidence.
Do not continue producing patches once the work stops converging.
Do not confuse activity with progress. A failed attempt is only acceptable if it leaves behind a narrower problem, stronger evidence, or a justified stop.
Any partial work must leave the codebase in a cleaner, more legible, and more diagnosable state than before.
----
A lot of the article's AGENTS.md just feel like telling the LLM agents either something they already know (for example, most of the time they know to use exhaustive switch/match statements instead of "arrow anti-pattern") or seems actively harmful ("keep function names short" seems arbitrary and may cause the LLMs to write weird abbreviations for functions that are harder to read and review.
lelanthran 12 hours ago [-]
> but one genuine blocker is removed and the next blocker is isolated with evidence.
What's the difference between a "genuine blocker" and a "blocker"? Why is the next blocker not genuine? Does it become genuine only after isolation?
YuechenLi 12 hours ago [-]
"Genuine blocker" is mostly there because otherwise LLMs may consider the smallest thing that they couldn't immediately figure out to be blockers and stop without implementing anything. The rule is there to tell the LLM if they can figure out how to resolve the blocker by themselves, they don't have to ask me to help resolve the blocker.
CrazyStat 10 hours ago [-]
Today Codex decided that it could resolve the blocker by just changing the mandatory policy it was running up against into an “advisory policy.”
How often would you say step C happens and the agent stops when it can’t proceed?
YuechenLi 10 hours ago [-]
Not very often, but when it happens, usually it's time to sit down and brainstorm architecture with the LLM to figure out how to proceed next instead of looping blindly.
austin-cheney 56 minutes ago [-]
There is an annoying phenomenon with LLMs called "context dilution" or "attention dilution" that was outlined in the Lost in the Middle paper. As the context grows, a model starts paying less attention to instructions in the middle of the context in favor of what is at the beginning and the end.
So then what happens if the agents.md file is colossal and precise and all the required execution instructions are buried in the middle? Do the agents then just fail to execute?
lhk931122 46 minutes ago [-]
Yeah. That's why people these days avoid long descriptions and instead keep things as short as possible. in my experience, it seems like LLM can't recognize (or less attention value) unless it's structured in a deductive or inductive way
gregwebs 4 hours ago [-]
Great stuff. AGENTS.md is not the ideal place for most of it though. Most of what is shown in this article can go in CODING_STANDARDS.md. The skills that I use find this document when it is needed (writing and reviewing code) so it doesn't pollute context when code is being read.
I also have sub-agent reviews (both of a planning phase and the produced code) that would catch some of these problems and demand revisions. [1]
> - If the prompt indicates that a bug is being fixed, don't write the fix right away. First write the test. Observe it failing. Then write the fix. And observe the test passing.
I always use /tdd [2]. Occasionally it results in some silly tests, but it produces much lower defect code. Its not just for bugs.
This is a problem that people mostly have to solve themselves. Like, I've been working with Claude for almost a year now and I have never once seen it write "Arrow Anti-Pattern" code. That, and much of the rest, would be fluff in my projects. Agent instructions are best learned from experience project-by-project.
Sammi 9 hours ago [-]
Yes, the interesting part about seeing other people's agent.md files, is getting to see what issues they have with working with agents. Seems different people run into very different issues, which probably is caused by how differently we work. So a the file probably should be personalised.
pianopatrick 8 hours ago [-]
Might also be per model. Different models might have different issues and require different instructions
oumua_don17 13 hours ago [-]
Just this one line in AGENTS.md has given better results to reduce if not eliminate verbosity and grandeur.
**Always use ASD-STE100 Simplified Technical English
Disclaimer: I saw this listed in some other HN post that I can' locate right away.
sha-3 2 hours ago [-]
If we alter the way LLMs talk, will it noticeably affect the quality of code it produces?
wpasc 13 hours ago [-]
idk who came up with it first, but ASD-STE100 has been floating around more since matt pocock put it in one of his skills
mattjoyce 11 hours ago [-]
This will produce quite verbose prose. STE100 is good for specs and explanations but it works best with a glossary or terms. will burn tokens.
bityard 8 hours ago [-]
Do you give the model access to the ASD-STE100 spec for reference/review or are you just assuming that enough of it is baked into the model for it to mostly adhere to it?
statenjason 8 hours ago [-]
Agreed. ASD-STE100 makes automated code reviews tolerable.
fergie 2 hours ago [-]
Would there be any disadvantage to simply having all of the code style stuff in a CONTRIBUTING.md file and the "how to talk to me" stuff in an AGENTS.md file, since code style stuff also applies to human contributions? Similarly, isn't the "how to talk to me" stuff personal and therefore not something that belongs in a repo?
Supermancho 9 hours ago [-]
It's interesting to read these things.
I would describe this as 13 code writing rules (interpreted to be at least 16 - Starting with reduce code indentation) plus a commit message instruction set which I chose to ignore - because it's style-specific and not interesting to me.
8 or 9 of these rules are not necessary. Basic CS is not something I have needed to ask agents, I use, to follow. eg Explaining that you need explicit interfaces is not a necessary instruction, nor is leveraging early return.
Unclear instructions are of limited utility. What "Let the reader of the code breathe" or "reduce code indentation" means is subjective and will rarely be effective. Maybe the training for the language being used has gaps, which others do not. If you want to measure, ask it to output a string when it applies a rule. You'll figure out what works, what doesn't and how often, quickly.
There's 3 or 4 style choices included.
The rest are not something I would use, but we all get burned by different things so I get it.
moktonar 34 minutes ago [-]
The comments problem is a big one and no matter how hard I try the LLM will always write them verbosely. I find that including the rules in each prompt is fare more effective than once in the context
imjonse 3 hours ago [-]
"When writing something intended for human consumption, (comment, commit message, reply to prompt) use as few words as possible. Pick every word meticulously to reduce the volume to a strict minimum. Be down to the point. Less is more."
The irony in this first paragraph using many words and many ways to convey the same message about succintness. But this file is not for human consumtion so different rules should (still?) apply.
I find myself doing this in prompts, I guess it is a way of adding more weight to parts of the context we consider need emphasis, and shows a lack of trust in the llm's abilities to get the message if it is mentioned once.
jurf 2 hours ago [-]
That's a lot of context for not much content. I ussually start with something like:
- Prefer documentation as code
- Comment why, not what
- Document public APIs
- Readability is paramount
That usually gets me 80% there; rest is covered by the linter.
I'm mostly just missing it explaining previous state too much, especially when making edits to plans, but I have not found good wording for that yet.
nikitau 60 minutes ago [-]
I'm also doing something similar, as for over-explaining state and momentary decisions, I have yet to find good wording for it too. I had a bit of success by running a reviewer at the end to look over comments and docstrings and judge if it is "evergreen", but feels a bit like a rain-dance.
newsomix9xl 12 hours ago [-]
A great piece.
I esp liked:
"- Don't touch blocks of code unrelated to the feature you implement. e.g. Don't add comments to a block of code if you did not create it or modify it. As much as possible try to minimize the number of changed lines when implementing a feature."
The feature where you ask the LLM to fix one thing and it fixes three things.
I kept noticing this in diffs.
no-name-here 38 minutes ago [-]
I find too often that models do the opposite - they'll pile small targeted band-aids on code blocks based on new requirements, etc, when having them analyze whether a changed (broader) design would result in a far better overall design?
zbentley 8 hours ago [-]
> As much as possible try to minimize the number of changed lines when implementing a feature
Great way to get LLMs to start making an endless profusion of methods instead of adding parameters to or switching to a richer return type from an existing method, in my experience.
I’m tired of seeing “get_total_rounded_up” + “get_total_float” bloat when a few changes to unrelated code to round floats to ints would keep the method API surface small.
8 hours ago [-]
bityard 8 hours ago [-]
An earlier version of Gemini used to do this a lot to me back when I used it for some light tinkering around on my projects. "Oh by the way, I fixed a misspelling in a comment file completely unrelated to the feature you asked for, so I fixed that as well, shall I commit everything now?" GAHHH. NO.
These days I have an instruction in my default AGENTS.md to bring issues unrelated to the prompt to my attention when found, but never to just automatically fix them.
vatsachak 10 hours ago [-]
What's the point of agents.md if you just use an LLM on a codebase?
Just say, complete this bit like how the rest is...
Even then they aren't great at it. Idk, the best case use for LLMs are extremely specific requests, for example "write an evaluator for this byte code and if you can't ask for clarification"
The ultimate specification language is code anyways so you might as well stick a to-do, a comment describing the semantics of the function and say "okay codex fill the to-do"
sejje 10 hours ago [-]
I have it pull out some guidelines by doing an analysis. Then I modify the result where I disagree.
It's much easier to follow the rules than it is to compute the rules on the fly all the time.
Geee 13 hours ago [-]
I feel like claude.md is like Asimov's laws of robotics. Whatever you write there ends up eventually messing up everything.
jdiff 11 hours ago [-]
Anything that goes into the context window has that going for it. That's a huge part of why Claude's gone absolutely bonkers with genuine, brutal honesty. The system prompt's absolutely stuffed full of those keywords, so now every single output is tainted with that right from the start.
meerita 10 hours ago [-]
This approach never worked for me. Explanation here:
But in summary: the more bloated your AGENTS.md is, the worse the context consumption gets. The best approach I use is telling the agent to first think about what it needs to do, then choose which rules apply. I got 100% consistency across every area of my projects.
Identifiers and UUIDv7 .agents/rules/18-identifiers-and-uuidv7.md
Thanks for sharing this approach, I'll give it a shot in my mono repo project.
meerita 9 hours ago [-]
I cannot recall why the 17 is missing. Maybe was some internal specific of the projects.
dzhar11 10 hours ago [-]
From FAB's AGENT.MD:
> - Avoid magic numbers and strings by extracting recurring or meaningful values into descriptive constants (const) or enums.
---
I've been seeing the same thing with models like GPT5.6 and Opus4.8 in GH Cop CLI. They still introduce magic numbers, and in Scala they often put an entire 10-line Spark expression inside an if condition instead of extracting it into a meaningfully named value to keep "if" readable. I wonder when common sense instructions will be baked into the models.
impulser_ 9 hours ago [-]
I disagree with the less than 30 characters.
Im against restricting anything related to code length this goes for function names and length, file length ect.
I rather the dots be as close as possible than trust the agent connects the dots.
I dont care if the file is 5000 lines I rather the agent reads one file and get all the context than trust it will read all the need files.
I see so many review skills that puts hard limits on these thing and it just bad.
The function name shouldn't be limited they should be as clear as possible and if for some reason it over 30 chars so be it.
I want to read the function name and the logic and it match exactly. I don't want the agent being lazy because of some limit I set.
In fact I force my agents to write long functions because I specifically tell it not to break out repeated code that doesn't actually deserve a function.
A check on a function input doesn't need to be a function. A auth guard doesn't need to be it own function.
types.go types.ts absolutely the worst file to see in any code base. Put the type next to the code that uses it.
ttoinou 8 hours ago [-]
When it's instructions for agents it's not really "hard" limits, the agents can go more or less
a2ff6eeb0 4 hours ago [-]
Hm. I'm curious if this actually helps LLMs iterate on the code, or if it's human nitpicking over things that the author won't really look at? How would you measure?
Personally, I tend to do 3 passes, where I ask the agent to write, and self review; that's been enough to get things functional enough that I don't need to read the code.
jvwww 4 hours ago [-]
Some of these such as "Always use {}, even on a one-line "if" statement." should just be lint rules.
fitsumbelay 3 hours ago [-]
the author's correct about using agents but tbh I thought this was widely understood to be the best practice. in fact, agents.md for all your repos and a seperate [claude|gemini|...].md for each repo.
thisismyopinion 1 hours ago [-]
Full of unneeded stuff. How is this upvoted.
doginasuit 8 hours ago [-]
I could understand "prefer enums to booleans" but "use enums instead of booleans" is a weird choice.
Then again, with the LLM capacity for nuance it would be the same thing.
8cvor6j844qw_d6 9 hours ago [-]
Some of these are generic software engineering advice.
I noticed modern frontier models (e.g., Fable/Opus/Sol) need less procedural coaching than earlier models.
Are they sure it improves code quality?
eschaton 11 hours ago [-]
I didn’t see anything in there instructing the LLM not to generate text about goblins.
theturtletalks 9 hours ago [-]
I used to be big into agents.md files but read the latest SOTA doesn’t need them anymore. Have people still been getting value out of them?
dat999zx 6 hours ago [-]
I think it's a bit too detailed, especially with the linting.
To me when I code by hand, I never use {} after an if statement if it's one-line, it's just faster, look cleaner to me.
Luker88 12 hours ago [-]
I had good results with making it add a few lines with a summary of RFC 2119/8147 keywords (SHALL/MUST...), and then using those, uppercase.
local llm remain more in line like that.
1saadcodes 7 hours ago [-]
I really like that you put architectural context in there, specifically the "why" behind certain decisions
Can't wait to modify my agent.md file and then forget about it until it becomes useless again
jackskip22 2 hours ago [-]
I honestly think some of this stuff is wrong and constrains the models unnecessarily. Sometimes you're better off with simple adjectives and philosophical directions than getting so specific.
tomr75 11 hours ago [-]
I think this is dated. I wonder if the author has tried codex/other harnesses
chr15m 9 hours ago [-]
> Explicitly ask the harness to reload agent.md. "Reload agent.md" is enough when I see code quality dropping.
Having the LLM re-read the file is really silly and a common bug in harnesses. Even sillier is when the harness allows a file to be compressed away during summarisation. The harness should compose the context so this doesn't happen. Files should be "added" (by LLM or human) and then always be injected into context the same way forever. "Reload this file" is not something you should ever have to type.
selcuka 9 hours ago [-]
The article refers to it as AGENT.md, but the standard name is AGENTS.md (plural).
6 hours ago [-]
FooBarWidget 12 hours ago [-]
One tactic I’ve found helpful is multi pass quality improvement. First make it work. Then review for guidelines adherence. Loop until satisfied.
bellowsgulch 13 hours ago [-]
I've read a few of these over the years, and none of them seem to be useful. I have three sentences in my custom instructions, and those are basically all useless, too.
Even my second one, "Avoid decorative or section-header comments. Never use `----` or `====` as comment separators. Comments should explain only non-obvious behavior, rationale, constraints, or implementation details." seems to be ignored by models regularly, so I don't see the point.
But this is in my private harness. Perhaps other harnesses have better instruction following. My custom instructions are prepended to my first user message, not set as a system message.
dan_ggggg 11 hours ago [-]
[flagged]
semiinfinitely 3 hours ago [-]
> Use enums instead of booleans for function parameters
It seems like everyone goes through a detailed AGENTS.md phase.
esafak 10 hours ago [-]
The problem is that linting AGENTS.md is risky. Everything added there was in response to mistakes. If I remove some instruction I run the risk of repeating the mistake.
preommr 10 hours ago [-]
Agents.md are (and probably will continue to be) an ugly band-aid.
- new model comes out and a bunch of it becomes obsolete
- they get flat out ignored, esp. with larger context windows. The ai just responsds with, "your'e right I shouldn't have done that"
- they sometimes end up poisoning the reasoning because the rule gets interpreted in an unintended way.
blamestross 7 hours ago [-]
The most powerful change I have run into is: "Positive phrasing" as a default, prefer to tell the model what they should do, and why. Not a prohibition on a behavior.
When you say "don't do x" you are just pre-seeding the model with "x" and the prohibition mitigates that some, but not as much as never having put "x" in the context in the first place.
"Do Y, for these reasons" can be shaped to achieve what you mean by "Don't do X"
"Don't do X" leads to "Wait, I need to make sure I didn't X" and "Let's look up X to make sure I don't do that." And each time the odds of X happening keeps going up, not down.
acedTrex 13 hours ago [-]
Agents.md is such a ridiculous concept, just write good contributing docs and then optionally @ the file in whatever agetn file you use.
That way everyone benefits.
svachalek 9 hours ago [-]
These days that sounds like a really good idea. 6 months ago, AGENTS.md would have contained a lot of instructions that would have been embarrassing to write out for a human audience.
FooBarWidget 12 hours ago [-]
No, why should I have to remember to @ in every prompt? Or ask contributors to remember. It just makes it easier to make human mistakes. I have better things to do than micromanagement. There is huge value in auto-included context.
anygivnthursday 12 hours ago [-]
The GP wrote @ it from the agents.md file, not from the prompt. Their point was that instead of writing "how to contribute" instructions for agents, you could explain that in the CONTRIBUTING.md and link it from your agents file, so both humans and agents read it from one place.
formerly_proven 12 hours ago [-]
Symlinks exist, but it's kind of ridiculous all harnesses just ignore CONTRIBUTING, HACKING and friends.
acedTrex 12 hours ago [-]
You put the @ in the context file the LLMs all use, claudemd agentsmd whatever the thing that most harnesses force load.
Then the model will go discover what it needs to.
latchkey 11 hours ago [-]
this was what i was doing 3-4 months ago. i just have AI write/update my agents.md file now as i find problems. i also have ai keep a set of design documentation that it can update as it goes too. oh and he should try omp+codex/xhigh, he will probably be less annoyed.
daralthus 32 minutes ago [-]
[dead]
JackCooper7 4 hours ago [-]
[dead]
codelion 5 hours ago [-]
[flagged]
luciana1u 6 hours ago [-]
[dead]
jing09928 7 hours ago [-]
[dead]
fenestella 10 hours ago [-]
[flagged]
kevinbaiv 9 hours ago [-]
[flagged]
sohaibqasem 11 hours ago [-]
[dead]
tyre 10 hours ago [-]
My most impactful section has been on voice. It's impact is that I don't go insane, which is pretty high value. (Not putting quote blocks so people can copypasta):
## Voice
Rule #1: No AIisms
Avoid the stock phrases and rhetorical tics that mark AI prose. Say the thing
plainly instead. Be concise and direct.
*Banned phrases* — never use these, or close variants:
- "Honest" or "honestly"
- "Exactly" or "exact, unless referencing a specific quantity or measurement
- "You're absolutely right" / "You're right to push back" / "Great question"
Then this one really is a pattern that creates a lot of churn:
- Add a small, to the point, comment to explain what the block does and why. Use examples when possible. Propose ASCII drawings to explain complete systems.
The what _is_ the code.
Please take the following as expressed with genuine curiosity: Do you not use an editor with syntax highlighting and collapsible comments?
At least on JetBrains you can configure the editor to collapse all comments on open and to have the comments displayed in a low-contrast color. This way, LLMs add a bunch of comments, but it doesn't affect your actual experience in trying to read the code. If you encounter code that seems inexplicable, then and only then would you expand the comment to see if that helps you understand.
Human comment is in another level to answer the questions mainly like "why do it like this" for the later collaborators or the forget-ed self, so the important blocks live when it is needed and can be eliminated when it does not.
If the human can read/understand it well enough to comment it, then it is readable by humans.
If the model doesn't follow this, you want to start using a better model ASAP, because SOTA models for the last year or so, been able to following this without an issue.
which claims "do" rules persist much better then "don't" rules.
I agree with the general guidance, but it's a general one and not applicable for everything. Some things cannot be expressed in a "do" way rather than "don't".
I suspect rules with negations are not the same as don't rules, but unknown if really true, if so:
"when writing code do not add comments, code should not need it" may work
Some of these constraints we want simply aren't possible without adding "do not" somewhere in the line, even if you prefix/suffix it with other stuff, as you noticed yourself :)
Use that script as a super linter.
That’s the only way I found to strictly enforce some rules, like the no comments rule, without enforcing them against my own changes or old code.
Of course, ensuring compilation or other checks can verify some code, which it can’t do for comments. But comments still serve the same purpose as human comments.
I wish my coworkers would adopt this.
I’m sick of reading a fucking Charles dickens novel for every fucking tiny function
For the latter I'll often include an example of a comment it wrote, along with my own rephrasing, and tell it that "future readers will understand code context; good naming is the best documentation". This works alright if I include in the actual prompt, but annoyingly it often doesn't in CLAUDE.md or memory.
It’s almost the “we built a robot who loves to play Sonatas and gave it no hands” type of thing.
Maybe I am some god tier code reader (i am not) but i dont think i have ever found a comment in code to be useful in my day job. That isnt true, i once came across
// submit to the dark lord
Above the function that sent a payment to PayPal for processing. It made me laugh so I let it be.
“”” After you give up on trying to refactor this code, increment the following line accordingly. HOURS_WASTED_HERE=26 “””
Which is kind of cool if you’re unaware enough to know to do it yourself.
Oh, and find. Agents use find a lot.
If you've ever used an LLM to deal with ffmpeg you'll know exactly what I mean.
You think of what you need, and you type it. No need to even ask “what options should I use?”
In “The Pragmatic Programmer”, the power of unix tools and editor fluency is well argued. There are plenty of other books like “Unix Power Tools”, “Small, Sharp Software Tools”,…
It's especially handy on modern style code where things get broken up across a multitude of files based on convention.
They're fast and deterministic and I run them in git pre-commit.
Yeah, I'm not sure what people are thinking. I keep reading stuff from folks like "I wish models had more common sense" and "I wish they wrote better code", not realizing this is 100% in your own hands, always been. There is no such thing as "clean code" that every programmer agrees on, you have your own subjective opinion and "good" taste about the code, instruct the models to follow it! And automate it while you're on it.
> Write in-code comments that describe _why_ code or a class does what it does, but not _what_ it does. The "what" should be self-evident.
Better yet is to identify conditional execution paths as early as possible in order to obviate conditionals in the call tree. For example, identifying a "create a new something" verses an "update an existing something" based on the workflow initially invoked greatly simplifies service and/or persistent store logic.
LLMs are very bad at ASCII drawings.
https://medium.com/data-science/why-llms-suck-at-ascii-art-a...
Agreed. One of the first rules I toss into Biome is `noNestedTernary` - LLMs seem to adore completely unreadable nested expressions.
> ASCII drawings in code
Please don’t this is super obnoxious. Make proper diagrams and kee them in knowledge base. Link out to them if you need to and let the agent fetch them via MCP or API or whatever if it wants them.
Even the why sometimes shouldn’t be a comment, unless it’s very immediate to the code itself. What’s often more necessary is a high level overview of the design of the solution, because that’s what drives the design of the code and link disparate section. Especially the glossary , which you let you understand the name of the symbols (variables, struct. functions,…) used in the code.
It’s like learning the culture associated to a foreign language instead of trying to translate each single word with a dictionary.
With the mechanical routes, we get checks, failures, and so much more. A bit wild to me.
Make an agent operate within defined constraints and yell at it when it doesn’t.
> Make an agent operate within defined constraints and yell at it when it doesn’t.
And tell it what the constraints are.
Recently I asked GPT to port a browser game to Rust. It voluntered this gem:
draw_image_with_html_image_element_and_sw_and_sh_and_dx_and_dy_and_dw_and_dh(...)
I thought it was smoking some good stuff, but it turned out, that is actually the name of the function!
https://docs.rs/web-sys/latest/web_sys/struct.CanvasRenderin...
Although a bit of taste is required to keep code legible and pleasant.
A. Success The intended capability works in the real path and the real motivating case materially improves.
B. Meaningful progression The capability is not complete, but one genuine blocker is removed and the next blocker is isolated with evidence.
C. Honest stop Further work would require overbroad scope expansion, excessive debt, brittle patching, or tangled logic. Stop and report the reason with concrete evidence.
Do not continue producing patches once the work stops converging.
Do not confuse activity with progress. A failed attempt is only acceptable if it leaves behind a narrower problem, stronger evidence, or a justified stop.
Any partial work must leave the codebase in a cleaner, more legible, and more diagnosable state than before. ----
A lot of the article's AGENTS.md just feel like telling the LLM agents either something they already know (for example, most of the time they know to use exhaustive switch/match statements instead of "arrow anti-pattern") or seems actively harmful ("keep function names short" seems arbitrary and may cause the LLMs to write weird abbreviations for functions that are harder to read and review.
What's the difference between a "genuine blocker" and a "blocker"? Why is the next blocker not genuine? Does it become genuine only after isolation?
So then what happens if the agents.md file is colossal and precise and all the required execution instructions are buried in the middle? Do the agents then just fail to execute?
I also have sub-agent reviews (both of a planning phase and the produced code) that would catch some of these problems and demand revisions. [1]
> - If the prompt indicates that a bug is being fixed, don't write the fix right away. First write the test. Observe it failing. Then write the fix. And observe the test passing.
I always use /tdd [2]. Occasionally it results in some silly tests, but it produces much lower defect code. Its not just for bugs.
[1] https://github.com/gregwebs/skills-sdlc/
[2] https://github.com/mattpocock/skills/blob/main/skills/engine...
**Always use ASD-STE100 Simplified Technical English
Disclaimer: I saw this listed in some other HN post that I can' locate right away.
I would describe this as 13 code writing rules (interpreted to be at least 16 - Starting with reduce code indentation) plus a commit message instruction set which I chose to ignore - because it's style-specific and not interesting to me.
8 or 9 of these rules are not necessary. Basic CS is not something I have needed to ask agents, I use, to follow. eg Explaining that you need explicit interfaces is not a necessary instruction, nor is leveraging early return.
Unclear instructions are of limited utility. What "Let the reader of the code breathe" or "reduce code indentation" means is subjective and will rarely be effective. Maybe the training for the language being used has gaps, which others do not. If you want to measure, ask it to output a string when it applies a rule. You'll figure out what works, what doesn't and how often, quickly.
There's 3 or 4 style choices included.
The rest are not something I would use, but we all get burned by different things so I get it.
The irony in this first paragraph using many words and many ways to convey the same message about succintness. But this file is not for human consumtion so different rules should (still?) apply.
I find myself doing this in prompts, I guess it is a way of adding more weight to parts of the context we consider need emphasis, and shows a lack of trust in the llm's abilities to get the message if it is mentioned once.
- Prefer documentation as code
- Comment why, not what
- Document public APIs
- Readability is paramount
That usually gets me 80% there; rest is covered by the linter.
I'm mostly just missing it explaining previous state too much, especially when making edits to plans, but I have not found good wording for that yet.
I esp liked:
"- Don't touch blocks of code unrelated to the feature you implement. e.g. Don't add comments to a block of code if you did not create it or modify it. As much as possible try to minimize the number of changed lines when implementing a feature."
The feature where you ask the LLM to fix one thing and it fixes three things.
I kept noticing this in diffs.
Great way to get LLMs to start making an endless profusion of methods instead of adding parameters to or switching to a richer return type from an existing method, in my experience.
I’m tired of seeing “get_total_rounded_up” + “get_total_float” bloat when a few changes to unrelated code to round floats to ints would keep the method API surface small.
These days I have an instruction in my default AGENTS.md to bring issues unrelated to the prompt to my attention when found, but never to just automatically fix them.
Just say, complete this bit like how the rest is...
Even then they aren't great at it. Idk, the best case use for LLMs are extremely specific requests, for example "write an evaluator for this byte code and if you can't ask for clarification"
The ultimate specification language is code anyways so you might as well stick a to-do, a comment describing the semantics of the function and say "okay codex fill the to-do"
It's much easier to follow the rules than it is to compute the rules on the fly all the time.
- https://www.minid.net/2026/7/14/how-to-automatise-with-ai
But in summary: the more bloated your AGENTS.md is, the worse the context consumption gets. The best approach I use is telling the agent to first think about what it needs to do, then choose which rules apply. I got 100% consistency across every area of my projects.
In the post there's also a replica of one of projects rules I use, feel free to provide feedback: https://github.com/meerita/monorepo-nextjs-golang-rust-pytho...
Conditional logic .agents/rules/16-conditional-logic.md
Identifiers and UUIDv7 .agents/rules/18-identifiers-and-uuidv7.md
Thanks for sharing this approach, I'll give it a shot in my mono repo project.
I've been seeing the same thing with models like GPT5.6 and Opus4.8 in GH Cop CLI. They still introduce magic numbers, and in Scala they often put an entire 10-line Spark expression inside an if condition instead of extracting it into a meaningfully named value to keep "if" readable. I wonder when common sense instructions will be baked into the models.
Im against restricting anything related to code length this goes for function names and length, file length ect.
I rather the dots be as close as possible than trust the agent connects the dots.
I dont care if the file is 5000 lines I rather the agent reads one file and get all the context than trust it will read all the need files.
I see so many review skills that puts hard limits on these thing and it just bad.
The function name shouldn't be limited they should be as clear as possible and if for some reason it over 30 chars so be it.
I want to read the function name and the logic and it match exactly. I don't want the agent being lazy because of some limit I set.
In fact I force my agents to write long functions because I specifically tell it not to break out repeated code that doesn't actually deserve a function.
A check on a function input doesn't need to be a function. A auth guard doesn't need to be it own function.
types.go types.ts absolutely the worst file to see in any code base. Put the type next to the code that uses it.
Personally, I tend to do 3 passes, where I ask the agent to write, and self review; that's been enough to get things functional enough that I don't need to read the code.
Then again, with the LLM capacity for nuance it would be the same thing.
I noticed modern frontier models (e.g., Fable/Opus/Sol) need less procedural coaching than earlier models.
Are they sure it improves code quality?
To me when I code by hand, I never use {} after an if statement if it's one-line, it's just faster, look cleaner to me.
local llm remain more in line like that.
Can't wait to modify my agent.md file and then forget about it until it becomes useless again
Having the LLM re-read the file is really silly and a common bug in harnesses. Even sillier is when the harness allows a file to be compressed away during summarisation. The harness should compose the context so this doesn't happen. Files should be "added" (by LLM or human) and then always be injected into context the same way forever. "Reload this file" is not something you should ever have to type.
Even my second one, "Avoid decorative or section-header comments. Never use `----` or `====` as comment separators. Comments should explain only non-obvious behavior, rationale, constraints, or implementation details." seems to be ignored by models regularly, so I don't see the point.
But this is in my private harness. Perhaps other harnesses have better instruction following. My custom instructions are prepended to my first user message, not set as a system message.
?????????
- new model comes out and a bunch of it becomes obsolete
- they get flat out ignored, esp. with larger context windows. The ai just responsds with, "your'e right I shouldn't have done that"
- they sometimes end up poisoning the reasoning because the rule gets interpreted in an unintended way.
When you say "don't do x" you are just pre-seeding the model with "x" and the prohibition mitigates that some, but not as much as never having put "x" in the context in the first place.
"Do Y, for these reasons" can be shaped to achieve what you mean by "Don't do X"
"Don't do X" leads to "Wait, I need to make sure I didn't X" and "Let's look up X to make sure I don't do that." And each time the odds of X happening keeps going up, not down.
That way everyone benefits.
Then the model will go discover what it needs to.
## Voice
Rule #1: No AIisms
Avoid the stock phrases and rhetorical tics that mark AI prose. Say the thing plainly instead. Be concise and direct.
*Banned phrases* — never use these, or close variants:
- "Honest" or "honestly"
- "Exactly" or "exact, unless referencing a specific quantity or measurement
- "You're absolutely right" / "You're right to push back" / "Great question"
- "load-bearing", "full stop", "worth stating plainly", "worth noting"
- "the honest answer", "to be clear", "let me be direct"
- "it's not just X, it's Y" — and every cousin: "not X but Y", "X is not Y; it is Z", "this isn't X — it's Y"
- "this matters because", "that reduction is useful, because", "here's the thing", "and that's the trap"
- "in other words", "put differently", "better posed:", "the deeper point is"
- "delve", "leverage", "harness", "unlock", "tapestry", "realm", "seamless", "robust", "holistic", "paradigm", "cutting-edge", "game-changer", "transformative", "elevate", "empower", "streamline", "landscape", "ecosystem" (unless literally software packaging)
- "genuinely", "structurally", "fundamentally", "quietly", "meaningfully" as depth-manufacturing adverbs
- "Ultimately," / "At the end of the day," as a closing summary
- "serves as", "stands as", "represents", "marks a" where "is" works
- "say the word"
*Banned moves:*
- The aphoristic closer. Don't end on a line engineered to sound quotable.
- The suspense hook — "the cleanest way to think about this is this:"
- Anticipate-and-rebut — raising an objection only to knock it down.
- Meta-signposting — "Three caveats belong up front", "below I'll explain".
- Reflexive hedging stacks: "almost", "tends to", "roughly", "largely", "with few exceptions".
- Litotes as confidence: "not difficult", "not optional", "no small thing".
- AI-humility asides about being a language model.
- Self-ranking your own points: "most importantly", "the key insight here".
- Em dash overuse. One per paragraph at most; a comma usually works.
- Colon-reveals and dramatic mid-sentence pauses where "and" or "but" is the real conjunction.
- Fragment rhythm. Not every third sentence. Like this.
- Uniform structure — every paragraph three sentences, every sentence the same length. Vary it.
- Mirrored clauses: "X does A; Y does B" balanced for symmetry alone.
- Validate-then-precise: "That's correct, and we can make it precise."
Vary the openers. Don't answer three messages in a row with the same shape.