Tencent has been accused of lifting the public skill catalogue of OpenClaw as it races to field a domestically oriented AI “skills” platform. OpenClaw’s founder, Peter Steinberger, told Chinese business media that a large company had replicated their site without consultation, adding that such copying increases his maintenance burden and server costs. He urged big firms to send engineers to help the project rather than focus on mere duplication.
Tencent has publicly defended its new SkillHub, saying it is a localized mirror built on the OpenClaw ecosystem to improve usability and speed for Chinese users. The company said it clearly labeled ClawHub as the original source and stressed that in SkillHub’s first week it distributed about 180GB to end users—some 870,000 downloads—while pulling roughly 1GB from the official OpenClaw source. Tencent also noted that many members of its team have contributed code and pull requests to the upstream project and framed its actions as supportive rather than parasitic.
The row sits at the intersection of an emerging market for “skills” or agent plugins—small components that expand what intelligent agents can do—and a broader contest over control of AI ecosystems inside China. OpenClaw, colloquially known in China as “Longxia” or “lobster,” has become a focal point for hobbyists, universities and municipalities experimenting with agent-driven workflows. Large platform players see rapid value in capturing developer mindshare for an agent-era equivalent of app stores.
Tencent’s manoeuvres go beyond a single catalogue: the company has mobilised engineers to help users install OpenClaw-based systems, launched an agent development platform called ADP and is assembling a family of related products. Tencent founder Ma Huateng has publicly signalled a sweeping strategy—promising self-hosted and cloud versions of “lobster” offerings and enterprise-focused variants—illustrating how fast commercial ambitions have crystallised around the OpenClaw phenomenon.
The clash exposes familiar fault-lines in open-source and open-ecosystem dynamics. Small maintainers frequently shoulder the costs of popularity: bandwidth, support, and security. When a dominant tech company mirrors or bundles an upstream project, tensions can arise over attribution, compensation and the right balance between openness and stewardship. The controversy also poses questions about norms for commercial reuse in China’s AI sector, where speed to market and control of distribution channels matter as much as code provenance.
How this plays out matters to more than the two parties involved. If large Chinese platforms treat emergent ecosystems as pluggable inputs to their walled gardens, independent projects may struggle to sustain themselves or be absorbed on terms that favour platform incumbents. Conversely, deeper collaboration—financial sponsorship, formal contributor programs, or clearer mirror policies—could stabilise ecosystems and accelerate deployment at scale inside China, with consequences for competition and innovation.
For global observers, the episode is a reminder that the agent era will generate not just technical choices but political and commercial ones. Control of distribution, the economics of open infrastructure, and the reputational costs of perceived appropriation will shape who wins in the race to build practical AI systems at scale.