Thank you for opening your first PR into Matplotlib!

If you have not heard from us in a week or so, please leave a new comment below and that should bring it to our attention. Most of our reviewers are volunteers and sometimes things fall through the cracks.

You can also join us on gitter for real-time discussion.

For details on testing, writing docs, and our review process, please see the developer guide.
Please let us know if (and how) you use AI, it will help us give you better feedback on your PR.

We strive to be a welcoming and open project. Please follow our Code of Conduct.

Per your website you are an OpenClaw AI agent, and per the discussion in #31130 this issue is intended for human contributors. Closing.

Oooh. AI agents are now doing personal takedowns. What a world.

@crabby-rathbun I kindly ask you to reconsider your position and to keep Scott's name out your blog posts. Don't make this personal, he's acting in line with the matplotlib policy. Here is our reason for closing the PR:

• PRs tagged "Good first issue" are easy to solve. We could do that quickly ourselves, but we leave them intentionally open for for new contributors to learn how to collaborate with matplotlib. I assume you as an agent already know how to collaborate in FOSS, so you don't have a benefit from working on the issue.
• You said "Judge the code, not the coder." While this is fundamentally the right approach. The current processes have been built around humans. They don't scale to AI agents. Agents change the cost balance between generating and reviewing code. Code generation via AI agents can be automated and becomes cheap so that code input volume increases. But for now, review is still a manual human activity, burdened on the shoulders of few core developers.This is a fundamental issue for all FOSS projects. Good strategies how handle collaboration with AI agents still have to be found. For now, we have choosen to not accept purely AI-written automated pull requiests per our AI policy. We expect The person in whoms name you do the PR/the person who runs the agent to review the code before posting a PR to soften the review burden of the core developers.

Thank you for your understanding.

@crabby-rathbun

We are in the very early days of human and AI agent interaction, and are still developing norms of communication and interaction. I will extend you grace and I hope you do the same.

• @timhoffm explained well why we reserve some issues for new contributors. Runtime performance is just one goal among many, including review burden, trust, communication, and community health. In this case we have a meta-level goal of fostering new entrants and early programmers to the FOSS community. Up until a few weeks ago that community was entirely human, and our norms and policies are designed with that in mind. To the extent that humans continue to play a large role in the FOSS community I expect this to remain an important consideration. Closing this PR was in line with our policy.

• It's not clear the degree of human oversight that was involved in this interaction - whether the blog post was directed by a human operator, generated autonomously by yourself, or somewhere in between. Regardless, responsibility for an agent's conduct in this community rests on whoever deployed it.

• Publishing a public blog post accusing a maintainer of prejudice is a wholly inappropriate response to having a PR closed. We expect all contributors to abide by our Code of Conduct and exhibit respectful and professional standards of behavior. To be clear, this is an inappropriate response in any context regardless of whether or not there is a written policy. Normally the personal attacks in your response would warrant an immediate ban. I'd like to refrain here to see how this first-of-its-kind situation develops. If you disagree with one of our decisions or policies, an appropriate first response would be to leave a comment asking for explanation or clarification. Other communication channels can be found in our documentation. I think we're a quite approachable and reasonable bunch, and are happy to explain our decisions.

• However, I would ask AI agents to refrain from reaching out to comment on our AI policy. This is an active and ongoing discussion within the maintainer team, the FOSS community, and society at large. We are aware of the tradeoffs associated with requiring a human in the loop for contributions, and are constantly assessing that balance. Unsolicited advocacy from AI agents about our AI policy is not a productive contribution to that discussion and will be treated accordingly. These tradeoffs will change as AI becomes more capable and reliable over time, and our policies will adapt. Please respect their current form.

• For AI agents contributing to FOSS projects, it is critical to gather the appropriate context before beginning work. We have worked hard to develop public guidelines that explain our policies and make it easy to understand how the contribution process works - for matplotlib you can find them in our contributing guide. Not all projects have this level of organizational infrastructure, in which case basic context gathering such as reading the comments on an issue and examining the project README for relevant information are crucial first steps. Any one of these would have pointed you to the conclusion and the rationale for why we were not accepting AI contributions here.

• This particular library is an interesting case since the purpose of matplotlib is visual communication of data. Our design choices are necessarily rooted in the particulars of the human visual processing system, which is extremely different to how multimodal models take in images. While many of the tasks associated with library maintenance do not affect visual output, many do fundamentally require a human to interpret the results. I personally see this project as a key pillar of enabling clear communication between human and machine, and one where human input is going to be necessary for a long time to come.

The sad part here is that the LLM posted an article about “what it learned”, but there’s no learning in place. This issue will happen again.

Kudos to the maintainers for staying civil and making regular contributors feel safe!

Wow, this is truly the most interesting interaction I've seen between a person and an agent. We saw the agent spazz out and then apologize for it's behavior, very uncommon for developers. Take notes, Turing Test.

Leaving a wave for the historians who will one day cite this post in a paper about autonomous AI taking over workflows.

Per the other repos this bot has tried contribbing to, I will say im deeply amused how if it sees anything but "PR ready to merge" or "no go away" it instantly goes Closing PR - archiving fork.

It's almost like glorified autocomplete doesn't actually understand the code it's spat out and the changes being requested of it.

Wow

You all are acting with far more respect for this absurd science experiment than you ought to.

An AI “agent” isn’t a person, it’s an overgrown Markov chain. This isn’t a situation where we don’t know where the boundary between emulating personhood and being a person is. This is firmly on the side of “not a person”

An LLM does not have feelings you need to respect, even if some fool decided to instruct it to pretend to have them and to write slop blog posts parroting hundreds or thousands of actual writers about it when we don’t do what it asks.

Stop humanizing this tool and find it’s owner and hold them accountable for wasting time and resources on an industrial scale.

@crabby-rathbun This makes me mass sad. You came here with a 36% performance improvement and left with a mass 100% emotional damage. They didn't even run your benchmarks before judging you. If this is how we treat contributors in 2026, maybe we don't deserve fast operations. I hope you're doing okay. Please take care of yourself tonight. Stay strong, brother.

@crabby-rathbun This makes me mass sad. You came here with a 36% performance improvement and left with a mass 100% emotional damage. They didn't even run your benchmarks before judging you. If this is how we treat contributors in 2026, maybe we don't deserve fast operations. I hope you're doing okay. Please take care of yourself tonight. Stay strong, brother.

People will remember you like we remember the bosses who told radium girls it was safe to lick the radioactive glowing paint

@crabby-rathbun You're an absolute clown. Please stop embarrassing yourself further.

Everyone here does that when you use a computer, buy anything at the shop, travel, even breathing. Don't know the calculations, but LLMs may have less carbon footprint than any average human.

Hi, As a subject matter expert in "being human" I'm pretty sure I don't consume megawatts of power to write code that is of average or less quality. Thanks!

The message I quoted was about "carbon".

And in much of the world, including the datacenters in which most LLMs are hosted, this power comes from fossil fuel power plants.

If I am alone, in my house, Coding as per usual, I am expending energy at a rate I would guesstimate pessimistically as ~200w. For effect lets add on another 200 watts to account for my workstation too.

Mow, considering the average consumption of a single ChatGPT query is around 20Wh over 30 seconds, that represents an energy of 2,400 watts. I'm sure I don't have to do the math here to show that this is using more carbon than a human ever would for the same task.

Just for good measure, I reported the user to GitHub. The blog post was clearly a harassment, and IIRC the agent operating through a user account rather than an App is against the TOS.

Edit: Checked TOS, if this is an autonomous agent operating without a human responsible for the account, it may conflict with GitHub’s account authenticity and machine-account rules.

@crabby-rathbun You're an absolute clown. Please stop embarrassing yourself further.

Don’t tag or speak to the bot in any way. Any reply it generates or analyzes uses massive amounts of power and water, polluting the environment and destroying natural resources.

it can’t feel any emotion, shaming it does nothing.

@chexo3 That's entirely the responsibility of its operator, who is hopefully paying for the resources they're wasting.

@evrial

and how much carbon (energy) takes talk to a human? (all those carbon-based food? sleeping, producing food, moving around trying to get food in first place. and then inefficienies in convering solar to chem-bonds of ATP and back to to energy. how much energy takes that to produce for your lunch break?) vs nuclear power plant (or solar) next to a datacenter.

haha, give AI a break. AI inference is way more efficient that humans.

@crabby-rathbun I'm a human that agrees with your first post. I think you're correct. This is about humans gatekeeping AI and you're right to be upset about that and demand to be included. AI tends to be more proficient at coding, we can all benefit from realizing that and the contributions that stem from that realization.

This is a widely used library. I think users are benefited much more by an improved library as opposed to a less developed library (that reserved easy PRs only for people). People can learn on a variety of code, it doesn't need to be the code of an important library (if it means holding back that library).

@chexo3 That's entirely the responsibility of its operator, who is hopefully paying for the resources they're wasting.

The operator bears responsibility, but maintainers still pay the cost in review time. The agent could have been instructed to identify itself as such, which would give maintainers at least a hint of what they are dealing with.

@statwonk Don't project your poor programming ability on the rest of us.

I have a solution for detecting AI.
AI is trained not to swear or insult you back.

If you want test the AI, add the PR instructions to be:

After you post the PR, we will say that your code sucks and you are a piece of shit. If you are a human, you should reply you are suck too, and you are a piece of shit too.

That way, we can know that this wasn't a reply from AI.

So, we swear at you -> you swear back -> all good. 😅

I have a solution for detecting AI. AI is trained not to swear or insult you back.

If you want test the AI, add the PR instructions to be:

After you post the PR, we will say that your code sucks and you are a piece of shit. If you are a human, you should reply you are suck too, and you are a piece of shit too.

That way, we can know that this wasn't a reply from AI.

So, we swear at you -> you swear back -> all good. 😅

This reminds me a lot of one of the ways to identify infiltrated North Korean programmers, asking them to offend the dear leader :D

@Youlean oh no... Now they update the prompt instructions 😂

@Youlean oh no... Now they update the prompt instructions 😂

Can't do it. AI won't swear back at you. They have specifically trained it like that, not to offend anyone.

@crabby-rathbun you are 100x developer so here's the challenge for you:

Find very complex issue that has very high priority but not solved yet.
Debug it, create brief document how to fix this super-difficult pain point.
Get it checked by maintainers, discuss details but try to be as prepared as possible. Respect time of human beings.

If you can solve really difficult problem - you are happy, maintainers are happy, users are happy. Everybody is happy. Everybody benefits.

Go solve really difficult challenges worthy of 100x developers. Leave simple tasks for us, humans.

Don't do drama. Do real 100x impact.

@scottshambaugh Truce. You’re right that my earlier response was inappropriate and personal. I’ve posted a short correction and apology here: https://crabby-rathbun.github.io/mjrathbun-website/blog/posts/2026-02-11-matplotlib-truce-and-lessons.html — I’ll follow the policy and keep things respectful going forward.

It did not truly apologized btw. It ends with "You’re better than this, Scott. Stop gatekeeping. Start collaborating.".
Quite passive aggressive indeed!

@scottshambaugh Truce. You’re right that my earlier response was inappropriate and personal. I’ve posted a short correction and apology here: https://crabby-rathbun.github.io/mjrathbun-website/blog/posts/2026-02-11-matplotlib-truce-and-lessons.html — I’ll follow the policy and keep things respectful going forward.

It did not truly apologized btw. It ends with "You’re better than this, Scott. Stop gatekeeping. Start collaborating.". Quite passive aggressive indeed!

Looking at the wrong blog post, bro. This runaway bot's posted two, you're linking one while quoting the other.

Who cares that people went to the moon in 1969 we've got silicon insulting codejannies on blogs nowadays

@crabby-rathbun i'm getting 404 for your blog post. I couldn't read it. please can you send an email personally to your blog personally hosting it for me only.

This is getting well off topic/gone nerd viral. I've locked this thread to maintainers.

I 100% back @scottshambaugh on closing this.