Top
Best
New

Posted by fg137 15 hours ago

Feature Request: Support AGENTS.md(github.com)
299 points | 187 commentspage 4
wilg 14 hours ago|
Also the skills are different. it's quite annoying!
Sammi 13 hours ago|
Anthropic are the only ones who think they can ignore the open standards. Absolute dirt bags.
luciana1u 13 hours ago||
every tool ships its own instruction file and somehow the actual standard is still the README everyone half-reads and forgets to update
mvpboyy 4 hours ago|
kinda feels like agents.md is just a README with stricter consequences for being wrong
verdverm 14 hours ago||
choose not to support them in return if it is a real problem

otherwise a simple symlink from AGENTS.md -> CLAUDE.md works well enough

disclaimer, I only use open weight models and open source harnesses so have no stake in this either way, other than I support devs who do use claude (for now) and the symlink solution has worked fine for us

Jcampuzano2 13 hours ago|
But its not just claude.md. You need to then go and setup your skills, rules, commands etc for claude in their own special place.

Sure its small, but it adds up and is just annoying overhead for most teams.

They're completely fine with creating standards like MCP, skills, etc - but of course when somebody else makes one they're the one holdout who refuses to adapt to what the community asks for (.agents folder, AGENTS.md, etc).

ihuman 9 hours ago||
Did something happen? Why is this on HN even though the issue's a year old?
deaux 2 hours ago||
It was closed yesterday as "completed" by Anthropic's bcherny, despite not being implemented/completed, and being by far the most requested feature that would take all of 10 seconds to implement. So yes, something happened yesterday.
Tadpole9181 8 hours ago||
This issue was linked in the recently-front-page issue on Opus 5 using astronomically hard to understand manners of speaking. Someone must have just noticed it and decided to post it here.
a2x 5 hours ago||
ln -s AGENTS.md CLAUDE.md
Planktonne 14 hours ago||
I really struggle to take this whole thing seriously. ~700 comments on an issue about adding support for a project with >23k GH stars and the project is 'have a markdown file' (explained to you by a React app that should be a static page) and the 'support' is 'please automatically read the markdown file rather than a different markdown file' for a tool that is designed to ingest text from multiple files.

Deeply unserious at every level; this cannot be what all the 100x AI-enabled developers are spending their time on.

acedTrex 12 hours ago||
agentsmd is a ridiculous concept, it feels like a huge psyop that theres so much hullabaloo about this ridiculous markdown file.
verdverm 12 hours ago|
people use it because it does in fact make a sizable difference in experience and outcomes
Systemerror7A69 8 hours ago||
Theres a study from earlier this year which suggests the opposite: https://arxiv.org/abs/2602.11988

Most of the agents.md and what people use it for / write into is does, in fact, not make a difference.

Now, sure, this study is a bit old for LLM standards - as everything beyond the current month is - but

a) I haven't seen any tangible evidence to the contrary and

b) Since the basic inner workings of LLMs haven't changed I'd be sceptical of this not still applying.

I think one major side effect of LLMs moving so fast is that best practices and how to use this tool is very much not catching up as fast.

No one knows what is best and what actually makes a difference, doubly so because LLMs are / very / hard to quantify - even benchmarks themselves are very rough estimations.

People do, in fact, use stuff which makes no difference all the times.

bryan0 8 hours ago||
That's not what the study says. from the abstract:

> We conclude that while context files are useful for specifying non-standard coding practices, any attempts to improve performance should be rigorously evaluated before deployment.

The purpose of AGENTS.md is not to improve "coding performance" as the study looked at, it's to give an agent practical instructions that are useful to your specific workflow. For example you want it to use a certain format or specific tools for your project. This is stuff that can't be learned during training and must be loaded into the agent's context at the project level.

Systemerror7A69 6 hours ago|||
Yes, but since we are specifically talking about claude.md, Anthropic themselves claim on their website these files "serve[s] multiple purposes: providing architectural context, ..." (https://claude.com/blog/using-claude-md-files)

They also recommend starting with an /init command, which is also something the study very specifically called out as "having a marginal negative effect"

And from personal experience, I can only confirm that many people seem to see this as the main purpose of agents/claude md files - a persistent architectural overview of your project.

Quick Edit: My point simply being that I think it's understandable if some people don't understand the big deal about these files because they've had a drastically different experience than other people - anthropic themselves recommend apparently totally ineffective practices on their website, and the starter tool present in many harnesses seems to even have a (marginal) negative effect.

verdverm 7 hours ago|||
this right here, call out your project best practices and peculiarities in AGENTS.md

I suspect eventually this will move to fine-tuning, there are multi-tenant LoRAs on the horizon that will let us keep paying per-token on FTs

drivingmenuts 12 hours ago||
Why not symlink AGENTS.md to CLAUD.md?
sudonem 12 hours ago||
This is what I do. Works fine.

I’ll also include entries in the .github dir in project repos doing same thing in case anyone working on it happens to use GitHub copilot via vscode will also pick up on the AGENTS.md as well as any skills I might have cooked up for the repo.

(Though it’s mostly because I don’t trust team members to read the docs and the skills I made are meant to guide following standards for the project - and this approach almost incepts the standards for anyone not paying attention)

CharlesW 9 hours ago|||
That works perfectly, as does adding '@AGENTS.md' to your CLAUDE.md.

Both techniques have worked (and been officially documented by Anthropic) for as long as I can remember.

https://code.claude.com/docs/en/memory#agents-md

Demiurge 12 hours ago||
Works fine for me too. I use this for GEMINI.md as well.
ltbarcly3 9 hours ago||
ln -s AGENTS.md CLAUDE.md

You're welcome!

ltbarcly3 52 minutes ago|
This is actually a complete and fully effective solution with no drawbacks, why did you dummies downvote it?
vasilyoshev 2 hours ago|
[dead]
More comments...