Search papers, labs, and topics across Lattice.
Ben-Gurion University of the Negev
3
0
5
Forget generic chatbots – SecMate slashes cybersecurity troubleshooting failures by 40% simply by adding device-specific diagnostics.
Multi-agent code editing with structured failure feedback boosts task success by 17%, suggesting a promising path to more reliable LLM-driven code manipulation.
Code Execution MCPs slash token usage and latency in agent systems, but open a Pandora's Box of new attack vectors, demanding layered defenses.
