agentseal

Security toolkit for AI agents. Red-team prompts, detect MCP poisoning, scan skill files, trace toxic data flows. 225+ tests across 28 agents...