Results for “vulnerability-research”

Three-phase vulnerability chain analysis — parallel agents find individual weaknesses, then a synthesis step identifies which combinations escalate to critical impact. Inspired by Strix (usestrix/strix) "Graph of Agents" pentesting model.

Perform read-only web research with governed browser proposals for manual review.

Conduct authorized security assessments of LLM applications and AI agents, covering prompt injection, tool abuse, RAG exposure, memory poisoning, and model supply-chain risks.