Is MIT License Enough? The Open-Source AI Skills Debate After Tencent's Mirror

Tencent launched SkillHub on March 11, 2026, mirroring 25,000 AI skills from ClawHub with CDN acceleration, Chinese localization, and Tencent's own editorial curation layer on top. They built a major platform on the work of thousands of independent skill creators.

They were legally entitled to do every bit of it.

The skills on ClawHub are MIT-licensed. MIT allows anyone to use, modify, distribute, and build commercial products on top of the work without asking permission and without sharing modifications back. Tencent followed the rules exactly.

The debate that followed isn't about what Tencent did wrong. It's about whether the rules are right for this specific type of creative work.

What MIT Was Designed For

MIT license was written for source code. The reasoning behind its permissiveness made sense in the context it was designed for: software libraries and utilities where the primary concern was enabling reuse and collaboration without legal friction.

The theory: restrictive licenses slow adoption. Open licenses accelerate it. Faster adoption means more testing, more improvements, more contributors, and ultimately more value for everyone — including the original creators who benefit from the increased visibility and the derivative work that comes back into the ecosystem.

This theory has worked well for code. Linux, Node.js, React, and thousands of foundational software projects run on MIT or equivalent permissive licenses. The ecosystem effects have been real.

AI skills are not quite code. They're closer to documentation, methodology, or instructional content — creative work that teaches an agent how to behave in specific contexts. The work of writing a quality SKILL.md file is closer to writing a well-researched how-to guide than to writing a software library. The value is in the knowledge and structure, not in executable instructions that would be impractical to reproduce independently.

What Tencent's Mirror Changed

Before SkillHub, the ClawHub ecosystem operated on the implicit assumption that MIT license meant community use: individual developers installing skills, small teams building on top of them, independent curators creating filtered lists.

SkillHub changed the scale. A single corporate actor took the entire catalog, applied proprietary curation criteria, embedded it in a commercial infrastructure play, and used it to drive adoption of Tencent's own products and cloud services.

None of this is license violation. But the gap between what skill creators imagined when they chose MIT and what MIT actually allows in the hands of a major corporation is significant.

The Arguments for MIT Staying Right

The case for keeping MIT, made by a substantial portion of the community:

Aggregation validates the work. When Tencent builds infrastructure around your skills, it means the skills matter. Complaining that a large company built on your MIT-licensed work is like complaining that your open-source library got popular. The license was doing its job.

Restriction creates fragmentation. If ClawHub switched to a more restrictive license — even something like AGPL or a custom license requiring attribution and non-commercial restrictions for large organizations — the ecosystem would fragment. Some contributors would choose the restrictive license; others would fork to MIT. Independent curation projects would have to navigate multiple license regimes. The coherence of the registry would suffer.

Chinese developers benefit from SkillHub. The practical effect of Tencent's platform is that Chinese developers who couldn't reliably access ClawHub due to network restrictions now have a usable alternative. Skills get more installs. Creators get more visibility. The aggregate download numbers on SkillHub show millions of downloads that wouldn't have happened without CDN acceleration in China.

MIT is a promise, not just a rule. Skill creators who published under MIT made a choice. Retroactively objecting to one specific use of MIT — because the aggregator is large and commercial — is changing the terms after the fact. That breaks trust with everyone who built on top of the MIT promise.

The Arguments for a Different Approach

The case for rethinking, also made by a significant community faction:

MIT doesn't account for the labor involved. Writing a high-quality skill that works reliably, handles edge cases, and documents its own behavior is real work. Developers who put significant time into skills aren't writing code libraries that take seconds to install and run. They're creating instructional content that requires domain knowledge and testing. The implicit exchange of effort for visibility and community benefit shifts when a corporation extracts the value without the creator having any relationship with the users of their work.

Attribution is absent in SkillHub. The SkillHub platform doesn't prominently attribute individual skill creators. Users download skills through Tencent's interface without necessarily knowing who wrote them or how to find the original. MIT requires that copyright notices be preserved in distributions — but for AI skills used via a web interface, what "preserving copyright notices" means in practice is unclear.

Corporate aggregators change the incentive structure. When individual developers build on top of your MIT skills, they often contribute back through issues, PRs, and new skills that build on yours. Corporate aggregators extract value from the ecosystem and return nothing to it. The incentive for individual contribution weakens if the most visible outcomes are corporate capture rather than community flourishing.

Practical Alternatives Creators Are Considering

Several licensing models have come up in community discussions since SkillHub's launch.

Creative Commons Attribution (CC BY) explicitly requires attribution in a way that MIT does not clearly require for non-software work. Creators whose primary concern is recognition rather than restricting use could adopt CC BY for their skills without significantly limiting distribution.

Business Source License (BSL) converts to full open source after a time period — typically 4 years — while restricting commercial use by large organizations during the conversion period. This is the model HashiCorp and others have used for developer tools that face competition from commercial cloud providers.

Dual licensing offers the skill under MIT for non-commercial use and under a commercial license for enterprise or large-organization use. The threshold (typically defined by revenue or headcount) determines which license applies. This is administratively complex but increasingly common in developer tool ecosystems.

Custom "fair use" licenses modeled on the Commons Clause or similar addenda that restrict specific types of commercial use while preserving the open-source properties that matter for individual developers.

None of these is clearly right for the AI skills use case. Each involves tradeoffs that affect different stakeholder groups differently.

What Skill Creators Should Actually Do Right Now

If you're publishing skills on ClawHub, here's the practical reality:

Under MIT, anything Tencent or any other aggregator does with your skills is legal. You cannot retroactively change this for skills already published.

For new skills, you can choose any license. If broad adoption matters most to you, MIT remains the most permissive and the most likely to be respected by third-party tools and aggregators. If attribution matters, add a CC BY notice. If commercial restriction matters, explore BSL or dual licensing — but understand that this will reduce adoption by organizations that prefer clear open-source licenses.

The more important question is strategic: what outcome do you want for the skills you publish? If your goal is maximum installs and broad ecosystem contribution, MIT maximizes that. If your goal is credit and relationship with users, the ClawHub platform itself (with download counts and star ratings) gives you more visibility than license terms.

The Underlying Question

The Tencent mirror forced a question the ClawHub community had not had to confront directly: are AI skills more like code, or more like content?

Code licenses were designed for reuse and modification. Content licenses were designed for attribution and creative rights. AI skills exist in the uncomfortable space between the two — they're structured enough to function like code, human-authored enough to feel like creative work.

The answer to "is MIT enough?" depends entirely on which metaphor you accept.

The community doesn't have consensus yet. What it does have is a real precedent: a major corporation built a major platform on MIT-licensed AI skills without contributing back. How creators respond to that precedent will define the license norms for the next generation of AI skill ecosystems.

The conversation is worth having before the next Tencent shows up.