mcp-evals/dist/types.d.ts

GitHub Action for evaluating MCP server tool calls using LLM-based scoring

import { type LanguageModel } from 'ai';
export interface EvalResult {
    accuracy: number;
    completeness: number;
    relevance: number;
    clarity: number;
    reasoning: number;
    overall_comments: string;
}
export interface EvalFunction {
    name: string;
    description: string;
    run: (model: LanguageModel) => Promise<EvalResult>;
}
export interface EvalConfig {
    model: LanguageModel;
    evals: EvalFunction[];
}
export interface YamlModelConfig {
    provider: 'openai' | 'anthropic';
    name: string;
    api_key?: string;
}
export interface YamlEval {
    name: string;
    description: string;
    prompt: string;
    expected_result?: string;
}
export interface YamlEvalConfig {
    model?: YamlModelConfig;
    evals: YamlEval[];
}
//# sourceMappingURL=types.d.ts.map