Tencent's SkillHub platform, an AI skill community built on the OpenClaw ecosystem and managed by Tencent Lighthouse Cloud, has reached 100,000 listed AI skills and ten million monthly downloads. The community launched in March and now competes with ClawHub as the largest similar marketplace in China.
The platform's engineering team emphasized that the main challenge is not distribution, but the discovery layer, which identifies the twenty percent of skills that actually work while hiding the rest.
SkillHub's response to this problem is a five-dimensional rating system called TRACE. These five dimensions include Trust, Reliability, Architecture, Compatibility, and Empathy. Priority is given to Trust, as any skill that fails origin or compliance checks is not included in the catalog at all.
Reliability functions as a performance level, Architecture relates to the API surface, and Compatibility assesses the match between the upstream model and the downstream client. Empathy is a softer parameter that determines how well a skill reflects user intent.
This system represents the purest deviation from the standard content community model. Unlike traditional forums, which organize offerings by topics and tags, distribute attention through ranking, block low-quality content via moderation, and assess quality through feedback loops, SkillHub performs all these functions plus one more step.
The rating here is determined not just by the number of likes, but by an evaluation performed upon upload and updated based on download volume. This metric decides whether a skill remains visible in the catalog or is demoted in ranking.
The product strategy has an obvious background: Tencent has elevated SkillHub's strategy to a top priority. The platform is now positioned as the central hub for agent skills for the Chinese market, existing alongside ClawHub in the West. The OpenClaw ecosystem update released in August became relevant this week, and Tencent Lighthouse Cloud serves as the primary hosting channel for OpenClaw-compatible skill backends in the Chinese market.
The platform acts as a bridge, connecting the Chinese developer who created the skill draft with the paying corporate user, with TRACE acting as the quality routing layer in the middle. Furthermore, the labor economy is worth noting. The TRACE system allows SkillHub to accept skills from a wide range of authors, including hobbyists and beginners using AI, without diminishing the overall quality signal of the platform.
Quality control on the supply side is now statistical rather than editorial. A skill that produces incorrect results when using standard test queries is excluded. The platform is designed not as a hall of fame, but as a working catalog, and this structure is an intentional decision. The main lesson SkillHub conveys to the rest of the Chinese AI industry is that the next competitive layer in the agent stack is not the model itself, but the platform that organizes the entire array of skills around it.
The Chinese agent InAgent achieved a task success rate of 90.2% on the OSWorld platform, becoming the first computer-operating agent to exceed the 90% threshold. Furthermore, it demonstrated 100% success in executing system-level tasks. This marks a transition to a new arena of competition in artificial intelligence—Harness engineering.
On July 27, the Chinese agent InAgent topped the global OSWorld ranking with a task success rate of 90.2%, winning titles in both general and agent categories. It surpassed the best public results achieved by Meta, Anthropic, and OpenAI. OSWorld, presented at the NeurIPS 2024 conference by the University of Hong Kong, Carnegie Mellon, and the University of Waterloo, is a recognized industry standard because it defines the threshold at which agents move from experimental use to stable industrial deployment.
The OSWorld platform tests applications in real Ubuntu virtual machines using 361 tasks covering office operations, programming, user interface development, and system administration. The evaluation is performed automatically, without human intervention. Progress dynamics showed growth from 12.2% in 2024 to 22.0% and 72.6% in 2025, when the human level was first exceeded, followed by 83.6% in May 2026, and finally 90.2% for InAgent two months later.
Result details include application collaboration with a success rate of 84.7% across 93 tasks, nearly ten points higher than second place; GIMP interface processing, where 24 out of 26 tasks were successfully completed with a 92.3% rate; and system-level operations executed with a perfect 100% result across 24 tasks, including process management, access rights, and command line.
Experts believe that the key to success lies not in the size of the model itself, but in Harness engineering. Harness acts as an AI operating system, determining how tasks are decomposed, how tools are called, and how recovery from errors occurs. There is a shift in industry understanding from Prompt Engineering to Context Engineering, and then to Harness Engineering. CITIC Securities notes that as agents move towards complex and multi-agent tasks, the costs associated with state maintenance, error propagation, and tokens increase, meaning the strength of a model in a single step no longer guarantees stable result delivery.
Harness represents an intermediate layer that transforms intelligence into stable operation, expanding the potential market from approximately $200 billion to $1.5 trillion, allowing AI to take over the management of legacy systems without needing APIs. Research from Stanford and Tsinghua demonstrates that different Harness implementations can lead to performance differences ranging from 6% to 17% when using the same model, and the percentage of successful coding tasks can increase from about 5% to over 30%.
The developer of InAgent, Intelligence Indeed, founded in 2018, is a pioneer in Harness. CEO Sun Linjun noted that pure API integration entails exorbitant maintenance costs: industrial systems like Siemens require up to 200,000 yuan per interface, and decade-old systems often lack interfaces, while e-commerce APIs are constantly changing. The agent takes no side; it uses interfaces where they exist and switches to a mode simulating human actions in systems without interfaces, observing the screen and controlling the mouse.
One manufacturing client previously spent half a year connecting a new system; now, by instructing the agent to compile production data into a report, they receive the result in just ten minutes. The company bet on the Graphical User Interface (GUI) as the core capability, based on the assumption that Artificial General Intelligence (AGI) requires a multimodal breakthrough: only by perceiving and operating in the world like a human can AI understand it. Harness transforms AI from a supported tool into an independent colleague, and the OSWorld results show that Chinese companies are leading, as AI competition shifts from simply piling up models to ensuring the stability of engineering solutions.
