/** * Optimization Techniques * * Research-backed prompting techniques for improving AI response quality. * Based on peer-reviewed research from MBZUAI, Google DeepMind, and ICLR 2024. */ import type { TechniqueConfig } from "./types"; /** * Expert Persona technique * Research: Kong et al. (2023) - 24% → 84% accuracy improvement */ export declare const expertPersona: TechniqueConfig; /** * Reasoning Chain technique * Research: Yang et al. (2023, Google DeepMind OPRO) - 34% → 80% accuracy */ export declare const reasoningChain: TechniqueConfig; /** * Stakes Language technique * Research: Bsharat et al. (2023, MBZUAI) - Principle #6: +45% quality improvement */ export declare const stakesLanguage: TechniqueConfig; /** * Challenge Framing technique * Research: Li et al. (2023, ICLR 2024) - +115% improvement on hard tasks */ export declare const challengeFraming: TechniqueConfig; /** * Self-Evaluation technique * Research: Improves response calibration and identifies uncertainties */ export declare const selfEvaluation: TechniqueConfig; /** * Analysis step (always included as first step) */ export declare const analysisStep: TechniqueConfig; /** * All available techniques */ export declare const ALL_TECHNIQUES: TechniqueConfig[]; /** * Get technique by ID */ export declare function getTechniqueById(id: string): TechniqueConfig | undefined; /** * Get applicable techniques for given complexity */ export declare function getTechniquesForComplexity(complexity: "simple" | "medium" | "complex"): TechniqueConfig[];