/**
 * Agent Benchmark Database — Store and compare benchmarks over time.
 *
 * Provides persistent storage for benchmark results with trend analysis,
 * run comparison, and dashboard reporting.
 */
export interface BenchmarkResult {
    /** Unique run ID (auto-generated if omitted) */
    runId?: string;
    /** Test or suite name */
    testName: string;
    /** Timestamp (ISO string, auto-set if omitted) */
    timestamp?: string;
    /** Pass/fail */
    passed: boolean;
    /** Duration in ms */
    duration_ms: number;
    /** Token usage */
    tokens?: {
        input?: number;
        output?: number;
        total?: number;
    };
    /** Estimated cost in USD */
    cost_usd?: number;
    /** Number of steps */
    steps?: number;
    /** Tool calls made */
    tools_called?: string[];
    /** Model used */
    model?: string;
    /** Arbitrary tags */
    tags?: string[];
    /** Custom metrics */
    metrics?: Record<string, number>;
}
export interface StoredBenchmark extends BenchmarkResult {
    runId: string;
    timestamp: string;
}
export interface TrendPoint {
    timestamp: string;
    runId: string;
    duration_ms: number;
    passed: boolean;
    tokens_total?: number;
    cost_usd?: number;
}
export interface TrendData {
    testName: string;
    points: TrendPoint[];
    avg_duration_ms: number;
    min_duration_ms: number;
    max_duration_ms: number;
    pass_rate: number;
    total_runs: number;
    trend_direction: 'improving' | 'degrading' | 'stable';
}
export interface ComparisonResult {
    run1: string;
    run2: string;
    tests: ComparisonEntry[];
    summary: {
        improved: number;
        degraded: number;
        unchanged: number;
        new_tests: number;
        removed_tests: number;
    };
}
export interface ComparisonEntry {
    testName: string;
    run1?: StoredBenchmark;
    run2?: StoredBenchmark;
    duration_change_ms?: number;
    duration_change_pct?: number;
    status_changed: boolean;
    verdict: 'improved' | 'degraded' | 'unchanged' | 'new' | 'removed';
}
export interface DashboardData {
    total_runs: number;
    total_tests: number;
    overall_pass_rate: number;
    avg_duration_ms: number;
    top_slowest: Array<{
        testName: string;
        avg_duration_ms: number;
    }>;
    top_flaky: Array<{
        testName: string;
        pass_rate: number;
        runs: number;
    }>;
    recent_runs: Array<{
        runId: string;
        timestamp: string;
        passed: number;
        failed: number;
    }>;
    cost_total_usd: number;
}
/**
 * Persistent benchmark database backed by a JSON file.
 */
export declare class BenchmarkDB {
    private dbPath;
    private data;
    constructor(dbPath?: string);
    private load;
    private save;
    /** Record a benchmark result. */
    record(benchmark: BenchmarkResult): StoredBenchmark;
    /** Record multiple results at once. */
    recordBatch(benchmarks: BenchmarkResult[]): StoredBenchmark[];
    /** Get trend data for a specific test over the last N days. */
    trend(testName: string, days?: number): TrendData;
    /** Compare two runs by runId. */
    compare(run1: string, run2: string): ComparisonResult;
    /** Generate dashboard data. */
    report(): DashboardData;
    /** Get all stored benchmarks. */
    getAll(): StoredBenchmark[];
    /** Get benchmarks for a specific run. */
    getRun(runId: string): StoredBenchmark[];
    /** List all unique run IDs. */
    listRuns(): string[];
    /** Clear all data. */
    clear(): void;
}
/** Format a benchmark comparison for display. */
export declare function formatComparison(result: ComparisonResult): string;
/** Format dashboard data for console display. */
export declare function formatDashboard(data: DashboardData): string;
//# sourceMappingURL=benchmark-db.d.ts.map