Tag: security
All the articles with the tag "security".
Pattern 9: Sandboxed Execution
A pattern for constraining agent-directed code execution with workspace, command, network, and approval policies.
Pattern 8: Memory Isolation
A pattern for making agent memory scoped, typed, validated, and isolated across users and agents.
Pattern 7: RAG Access Control and Provenance
A pattern for enforcing authorization during retrieval and carrying source provenance into generated answers.
Pattern 6: Prompt Injection and Goal Hijacking
A pattern for treating untrusted content as data and preserving the original user goal.