Your agent's 10,000-token system prompt, containing business logic, tool credentials, and RAG instructions, sits in a shared cache. A co-tenant on the same API endpoint can probe it token by token using nothing but response latency. Prompt caching reduces agentic API costs by 41-80%. That same feature creates a timing oracle: any co-tenant can... Weiterlesen
Intelligence View
⚡ tsecurity.de Intelligence
Prompt Caching Is a Timing Oracle: How the 41-80% Cost Win Becomes Cross-Tenant System Prompt Extraction
Your agent's 10,000-token system prompt, containing business logic, tool credentials, and RAG instructions, sits in a shared cache. A co-tenant on the same API endpoint can probe it token by token using nothing but response latency.…
Reagiere als Erste:r — dein Feedback zählt!
2. Cyber Threat Intelligence & Forensik
CTI Threat Relationship Graph2 Knoten / 1 Relationen