Framework for testing strategic reasoning in LLMs using evolutionary Iterated Prisoner’s Dilemma tournaments. Features 28 agents, variable horizons, memory conditions, high-concurrency execution, and detailed analysis of cooperation rates, strategic fingerprints, and LLM decision rationales across providers and temperatures.

Strategic foresight engine using IFTF methodology. Soft Predict (Claude-native) + Hard Predict (deterministic Python chain). Drivers, cross-impact, backcasting, four-scenario reports.