Xiaohang Yu 于晓航

~/sky/hacker/zero-days

• Founded and led Chaitin Tech’s blockchain security practice from zero (2018.03 to 2023.03), one of the earliest such teams in China.
• Led protocol and cryptographic security audits across major L1s: BTC, ETH, EOS, Cosmos and a dozen more chains; avg. 30 to 50k lines of Go/C++/Python per project, roughly 1M lines audited cumulatively.
• Around 100 zero-day vulnerabilities across consensus (PBFT, DPoS, PoC, VRF), P2P, and chain/block/tx verification layers, including novel vulnerability classes (Twin Tree and others); all reported and fixed. 5 CVEs from smart-contract audits (Solidity, Rust, C++).

~/sky/hacker/evm-jop

• EVM Jump-Oriented Programming, independent research, 2018.11 to 2019.02: first adaptation of JOP from classical binary exploitation to the EVM.
• Debuted as a Real World CTF 2018 Finals challenge; became a canonical, hardest-tier blockchain-CTF category.
• Presented at DEFCON 27 Village, Las Vegas, 2019 (“EVM Opcode JOP”).
• Also spoke at DEFCON China 1.0 Village, Beijing, 2019, on “Practical Security of Blockchain in Industry”, a separate talk from the JOP one, same year.

~/sky/hacker/standards

• Technical lead of the CNVD national Blockchain Vulnerability Scoring Standard, China’s first national blockchain vulnerability standard; coordinated 4 security vendors under CNVD/CNCERT guidance.
• Participated in 3 MIIT national industry-standard research projects: Blockchain Service Security Technical Requirement, Guidelines for Blockchain Smart Contract Security, Blockchain Security Monitoring System Technical Requirement.
• Authored the Blockchain Security Guidebook (2018), China’s first blockchain wikipedia, and the 2019 Blockchain Ecosystem Security Solution.

~/sky/hacker/ctf

• DEF CON CTF World Finals, 3rd place, team Tea Deliverers, 2021.
• Qiangwang Cup (national), first prize, 2020 and 2021.
• Co-organizer and challenge designer, Real World CTF blockchain track; also designed EVM-reversing and elliptic-curve crypto challenges for Huawei CTF, DVP CTF, and DoraHacks.

~/sky/hacker/origins

• OSCP (Offensive Security Certified Professional), 2017.12.
• Nights spent on pwnable.kr and VulnHub, written up on a study blog signed M1r4c13.

~/sky/builder/safeclaw

An agent that holds your API key holds your full authority. Every call, forever.

SafeClaw ends that. Agents only ever see phantom credentials; a local proxy injects the real secret at the network edge, and sensitive calls wait for a passkey tap. Each tap is a single-use grant bound to one exact operation, so even a fully compromised agent cannot spend your authority.

co-founded 2026 · production stack built solo in three months · open source

safeclaw.pro github the paper behind it

~/sky/builder/nodpay

• Also built NodPay.ai, an open-source wallet stack for AI-agent payments; prototyped in one week.

nodpay.ai

~/sky/builder/startup

• Technical co-founder of a blockchain infrastructure company, 2023 to 2025.
• Built and led an engineering team of 10+ (company 30+ at peak); owned technical direction, architecture, and delivery.
• Research and implementation of core protocols for verifiable AI systems and zero-knowledge-proof (ZKP) infrastructure.

~/sky/builder/open-standards

• Core technical contributor to open-source standards: opML (Optimistic Machine Learning), ERC-7007, ERC-7641.
• opML is optimistic ML verification, ERC-7007 covers onchain AI/AIGC assets, ERC-7641 is a fee-sharing token standard.

~/sky/scholar/education

• Ph.D. Computer Science, Imperial College London, UK, 2021.10 to 2027 (expected). Advisor: William Knottenbelt, also a LOCARD co-author.
• M.S. Security Informatics, Johns Hopkins University, US, 2016.08 to 2018.01. Fully funded by China Scholarship Council (CSC) national scholarship.
• B.S. Computer Science and Technology, Beijing Normal University, China, 2012.09 to 2016.06.
• National first prize, national undergraduate mathematical modeling contest, ranked 293 of 25,347 teams, 2014.12. ACM/ICPC Asia Regionals silver medal twice: Guangdong (92/798) and Mudanjiang (19/154), 2014.

~/sky/scholar/directions

• Agent security: cryptographic protocols for agent credential isolation and secret-use delegation (SUDP, first author).
• Agentic forensics: agentic frameworks for blockchain forensics (LOCARD, PhD work, IEEE ICBC 2026).
• Reinforcement learning: reinforcement-learning post-training for LLM reasoning (ACL 2026 main conference; TMLR 2025 agentic-RL survey, co-first author).

~/sky/scholar/publications

• SUDP: Secret-Use Delegation Protocol for Agentic Systems arXiv preprint, 2026 · first author
• LOCARD: An Agentic Framework for Blockchain Forensics code IEEE ICBC 2026 · first author
• Scaling Behaviors of LLM Reinforcement Learning Post-Training: An Empirical Study in Mathematical Reasoning ACL 2026 (main)
• The Landscape of Agentic Reinforcement Learning for LLMs: A Survey TMLR 2025 · co-first author
• earlier
• The Design of a Visual Tool for the Quick Customization of Virtual Characters in OSSL Yungang Wei, Xiaoran Qin, Xiaoye Tan, Xiaohang Yu, Bo Sun, Xiaoming Zhu · IEEE Cyberworlds 2015, pp. 314 to 320
• Exploring the Use of a 3D Virtual Environment in Chinese Cultural Transmission Yungang Wei, Xiaoye Tan, Xiaoran Qin, Xiaohang Yu, Bo Sun, Xiaoming Zhu · IEEE Cyberworlds 2014, pp. 345 to 350

~/sky/scholar/talks

• IEEE ICBC 2026 (upcoming) · LOCARD: An Agentic Framework for Blockchain Forensics · Brisbane, AU
• DEFCON 27 Village · EVM Opcode JOP · Las Vegas · 2019
• DEFCON China 1.0 Village · Practical Security of Blockchain in Industry · Beijing · 2019
• ISC Technical Forum · Beijing · 2020
• Yunqi 2050 · Hangzhou · 2018
• Real World CTF Forum · Beijing · 2019

9 talks, 2018 to 2026