What happened
Upwind Security announced (July 30, 2026) the AI Agent Context Scanner, which inspects agent skills, tools, MCP connections and instructions before execution for hidden/malicious prompts (mapped to OWASP LLM Top 10), alongside general availability of its AI DR runtime behavioral-baseline and anomaly-detection service.
Why it matters
Addresses the under-covered pre-execution 'context supply chain' (skills/tools/MCP instructions) as a distinct attack surface from the model itself, pairing it with GA runtime detection for a fuller agent-security lifecycle.
Applicability
Cloud-native security teams running autonomous AI agents in production; GA now for AI DR, Context Scanner newly available.