Weihang Niu
Titles as the sources gave them. In 1 entry, with 0 checked quotations.
In the Base
Every entry they appear in.
Scope-Gap Injection: Exploiting Evaluator Channel Blind Spots in the OpenClaw Agent Security Competition
According to the abstract, the authors describe an indirect prompt-injection method that hides bypass instructions in an attachment, so an LLM safety judge sees a harmless task while the agent follows the hidden list. They report results from an IJCAI 2026 agent-security competition and propose defenses.