AI Policy
Artificial intelligence coding agents have recently become very popular. Chariz continues to focus on supporting developers creating high-quality products, which requires us to define standards around acceptable use of artificial intelligence.
We understand that it is disappointing to receive a rejection. We don’t want to exclude anyone from being part of the Chariz seller community, however, we have long-standing seller guidelines outlining our expectations for quality and customer service. Inexperienced use of AI tools is very likely to hinder your ability to meet these expectations.
The situation around AI coding agents is continually evolving, so this policy may change frequently to reflect new changes. We will make our best efforts to contact sellers ahead of time, should their product(s) be in violation of upcoming policy changes.
This policy applies to revisions submitted on or after 24 August 2026.
Fully AI-generated projects
“Vibe coded” projects, which are fully developed from scratch by an AI coding agent with little to no code written by a human, will not be accepted.
Jailbreak tweak development is a niche that does not have strong representation in current AI models. In observing the use of AI in the community, we have found numerous examples of tools such as Theos, Logos, and Orion being used in unusual ways. This increases the difficulty of maintaining a tweak over time, whether by a human or agent.
Even with languages/ecosystems that agents are known to be good at working with, such as JavaScript, the code typically has the feel of a first draft by an inexperienced developer. Lack of programming experience makes it difficult to understand whether your coding agent is solving problems in a manner that is logical and likely to produce a satisfactory outcome, or if it is generating poorly-designed code that is likely to have performance, stability, and security issues. In some cases, agents are simply not able to solve flaws in code unless a skilled human operator is actively collaborating with it, to keep it on-track.
Code quality concerns are further heightened in the case of jailbreak tweaks, which can cause poor performance and crashes in operating system components. Frequent crashes can lead to more serious issues, such as corruption of user data.
Learning how to be a software developer allows you to understand when a coding agent is wasting time producing incorrect code, or poorly-structured code that is not conducive to long-term maintenance. Along with a growing amount of the wider hobbyist software development community, we recognise that the role of “software engineer” is far from obsolete. We strongly encourage anyone interested in this trade to invest time learning the skill, because the benefits to doing so are significant.
AI agent coding assistance
Chariz allows use of AI agents for assistance in developing a project. We consider this to mean that any generated code is reviewed and tidied up by a skilled developer. Consider whether you can leave the project for several months and return to it – do you believe you would be able to follow what the code is doing, without needing to ask an agent to explain it?
Use of AI agents for planning, such as in researching private APIs, triaging bugs and crash reports, and finding security issues, is acceptable. In fact, we have found these to be the strongest use cases for coding agents.
Marketing materials
Description/changelog text can be written with the assistance of AI, but should reviewed and tidied up by a human. If you are using AI tools because you are not a native English speaker, please let us know and we will do our best to help you write a description.
AI-generated assets (icons, banners, and videos) will not be accepted. In our experience, even with the best text-to-image/video models, AI-generated images very frequently contain errors. Even if there are no errors, users are frequently aware of the “tells” of an AI-generated image, and this will result in negative attention to your project. Such attention affects the reputation of Chariz and the other products on the platform, so we aren’t able to accept this.
Additionally, consider that we may suggest changes/improvements to any of your assets. Text-to-image/video models are not able to make changes in the same way they can be made to a hand-created image/video; this will add delays before your revision can be released.
Screenshots and screen recordings must accurately portray the product. As such, it is not acceptable for them to be generated by AI.
Disclosure of AI usage
Chariz is considering implementing a requirement to disclose whether AI tools have been used to develop or produce marketing materials for a product. This will appear in a sidebar section of the product’s listing page. We appreciate the community’s feedback on this matter, and we will continue monitoring how developers in our community use AI to determine the priority of this.
For open source projects, we recommend, but do not require, that commits containing AI-generated code be tagged as such in the commit message using a Co-authored-by: line. Most agents do this by default.
Trusted developers
We may make exceptions to our AI policies in cases where we are confident in the skills of a developer. Put another way, we will apply these policies more strictly to new developers who do not yet have a reputation in our community. This is because primarily AI-developed projects tend to become abandoned quickly.
Violations
Should we believe this policy has been violated by a submitted product, the revision will be rejected, and we will contact the seller regarding a solution that allows the revision to be approved. Please note that there may be cases where we do not see an immediate path forward, and may need to reject the product outright as it currently stands.
Should we believe this policy has been violated by an existing released product, we will contact the seller to discuss a solution. If we believe the seller has acted in bad faith, such as by not being upfront about the degree of AI usage in their product, we may consider removing the product.
In all cases, we will make our best efforts to allow the seller to respond, while also considering reputational concerns to our platform by approving the revision, or keeping the violating product available.