Security, Cost, and Production Operations | Agentic Engineering | Oh My Learning

01 The Threat Model: Why Agents Can't Reliably Separate Instructions from Data 12 min read mid drafted 02 Prompt Injection: Direct, Indirect, and Defense-in-Depth 13 min read senior drafted 03 Least-Privilege Tool Permissions and Scoped Credentials 14 min read senior drafted 04 Human-in-the-Loop Gates for Irreversible Actions 13 min read senior drafted 05 Token Economics: Model Routing, Caching, and Budget Design 15 min read senior drafted 06 Observability for Agents: Tracing, Logging, and Debugging Non-Determinism 14 min read staff drafted 07 Failure Modes and Verification: Hallucination, Silent Drift, and Trust Calibration 14 min read staff drafted 08 Operating Agent Fleets: Governance, Incident Response, and Organizational Risk 18 min read principal drafted 09 Agent Identity: Distinguishing Who Is Acting from Who Authorized It 17 min read senior drafted