ClawHub vs SkillHub: The AI Skill Marketplace Wars Have Begun
ClawHub and Tencent's SkillHub are now competing for the same developer audience. Here's how they differ on curation, licensing, business model, and where the ecosystem goes from here.
ClawHub vs SkillHub: The AI Skill Marketplace Wars Have Begun
Two registries now serve the same ecosystem. One is the original. One is backed by one of the largest technology companies in the world.
This is the story of how AI skill distribution is fracturing — and what it means for developers on both sides.
The Two Registries
ClawHub (clawhub.ai) is the official skill registry for OpenClaw, built and maintained by Peter Steinberger's team and now governed by the OpenClaw Foundation. It launched in mid-2025 and has grown to 13,729 skills with 1.5 million combined downloads. The backend runs on Convex. Search uses vector embeddings. Publishing requires a GitHub account and nothing else.
SkillHub (skillhub.tencent.com) launched March 11, 2026. Built by Tencent, it aggregates approximately 25,000 AI skills from the same underlying ClawHub ecosystem and serves them through Tencent Cloud's CDN infrastructure to Chinese developers. It's free. It has a curated TOP 50 ranking backed by Tencent's editorial team. It has no creator submission portal of its own.
The same skills. Different delivery. Very different stakes.
How They Differ
Curation Model
ClawHub's curation is community-driven. The registry itself applies minimal gatekeeping — anyone with a GitHub account can publish. Quality filtering happens in layers on top: the VoltAgent awesome-openclaw-skills list curates 5,366 skills; LeoYeAI's list filters to 339+ best-in-class, updated weekly; third-party sites like clawhub.biz and claw-hub.net maintain their own rankings.
SkillHub's curation is editorial and corporate. Tencent's team selects the TOP 50, applies triple security audits (security scan, content review, quality assessment), and controls what gets highlighted. There are no user ratings, no community stars, no feedback mechanisms. Quality flows from Tencent's judgment.
Neither model is obviously superior. Community curation is democratic but slow and inconsistent. Editorial curation is fast and consistent but reflects the curator's priorities, which may not match individual developer needs.
Business Model
ClawHub's business model is early-stage. The registry is free to use, free to publish on. OpenAI's sponsorship of the OpenClaw Foundation covers infrastructure costs. There's no current revenue model attached to the registry itself.
SkillHub is free and will remain free. Tencent's monetization is indirect: driving adoption of Tencent Cloud Lighthouse servers, promoting QClaw (their OpenClaw client for WeChat users), and positioning Tencent products as first-class citizens in the AI agent world through 10+ native Tencent skill integrations.
Geographic Reach
ClawHub reaches global developers, with natural bias toward English-speaking markets. International CDN delivery works fine for most developers but is slow in China.
SkillHub targets China explicitly. The entire interface is Chinese. The search handles Chinese natural language. The CDN is domestic. For Chinese developers, SkillHub solves a real problem that ClawHub couldn't solve without Tencent-scale infrastructure.
Creator Relationship
ClawHub is the only place you can publish a skill. If you want your work in the ecosystem, ClawHub is the source of truth.
SkillHub currently mirrors without ingestion. Skills on SkillHub got there because they were on ClawHub first. There is no separate SkillHub submission process. Creators publish once to ClawHub and get SkillHub distribution as a side effect.
The Licensing Debate
This is where the ecosystem's most interesting tension lives.
The OpenClaw project uses MIT license. ClawHub's codebase is MIT-licensed. Individual skills published to ClawHub carry whatever license their creators choose — many default to MIT.
MIT license means anyone can take, modify, distribute, and build commercial products on top of the work without permission and without contributing back. Tencent did exactly this. Their SkillHub is built on MIT-licensed code and MIT-licensed skills. They owe nothing under the license terms.
The community's reaction has been mixed.
Some developers see it as straightforward validation. When a company Tencent's size builds on your ecosystem without being legally required to, it means the ecosystem is worth something. The interest signals value.
Others argue that MIT is wrong for AI skills in ways it wasn't wrong for traditional open-source. When Tencent builds a mirror, they get to impose their own curation, their own security audit criteria, and their own priorities without the creator community having any input. A skill creator who publishes on ClawHub has no say in how Tencent presents, ranks, or potentially modifies their work for the Chinese market.
The debate is not resolved. What's clear: creators should understand that MIT-licensed skills are available for exactly this kind of aggregation. If you're not comfortable with that, you need a different license.
What the Competition Means for Developers
For developers building with ClawHub skills, the competition is entirely positive. More distribution means more installs, more feedback, and a larger test surface for quality skills.
For developers choosing where to discover and install skills, the choice depends on location. If you're in China, SkillHub's CDN acceleration and Chinese-language interface make it the practical choice. If you're anywhere else, ClawHub remains the primary registry with the most complete catalog and community-generated curation.
For skill creators, the implication is that publishing to ClawHub now delivers global distribution including China — without any additional effort. That's a meaningful expansion of your potential audience.
Where This Goes
The marketplace wars framing may be premature. ClawHub and SkillHub are not yet competing in any direct commercial sense. Neither charges users. Neither has locked-in business models that require the other to lose.
The more accurate framing: AI skill distribution is becoming infrastructure. When infrastructure matters enough, multiple parties build it. DNS has multiple providers. Package registries have regional mirrors. CDN providers compete on the same content.
The same pattern will apply to AI skills. Multiple discovery interfaces, multiple distribution mechanisms, multiple curation layers — all feeding from the same underlying registry.
The question for the ecosystem's long-term health is governance. The OpenClaw Foundation now oversees the base platform. If the foundation develops effective standards for skill quality, security, and attribution, those standards can flow through to SkillHub and any future mirrors. If not, the ecosystem fragments into incompatible dialects.
Tencent's launch made the governance question urgent. That might be the most important consequence of March 11, 2026.