Svmuu News: 0G Labs, in collaboration with research teams from the National University of Singapore, Peking University, and Beijing University of Posts and Telecommunications, has proposed the first automated testing framework, Agora, in a paper accepted for ICML 2026. The framework integrates domain knowledge with multi-agent collaboration using large language models, driving the automated testing architecture through the interaction of three specialized agents: the Coordinator, Strategist, and Code Officer. In tests of consensus protocol libraries such as Raft, EPaxos, HotStuff, and BullShark, Agora successfully identified 15 previously unknown deep protocol-level logical vulnerabilities. In contrast, native large language models such as GPT-5.2 and Claude 4.5 failed to detect these vulnerabilities. Evaluation results show that genuine logic flaws account for 73.9% of the vulnerability reports generated by Agora, with an average cost of approximately 5.32 million tokens—equivalent to about $40—per logic flaw discovered. The framework’s code has now been released to the open-source community.