The Tangent Rule: an opportunity-cost stopping controller for agentic AI (when to stop reasoning, retrieving, or looping). Reference code, regime-shift simulation, and a Claude Code plugin. Paper: Trivedi (2026).
reinforcement-learning mcp foraging ai-agents rag opportunity-cost optimal-stopping average-reward large-language-models llm agentic-ai test-time-compute claude-code inference-time-scaling marginal-value-theorem
-
Updated
Jun 23, 2026 - Python