Class AuditAnalyzer

Utility class for analyzing agent run data and applying smart truncation

Index

Constructors

Methods

calculateAggregateStats calculateStepDuration detectErrorPattern estimateCost estimateTokenCount formatDuration getTruncationRules suggestFixes truncateField

Constructors

constructor

new AuditAnalyzer(): AuditAnalyzer
Returns AuditAnalyzer

Methods

calculateAggregateStats

calculateAggregateStats(
    runs: { duration: number; success: boolean; tokenCount: number }[],
): {
    avgDuration: number;
    avgTokens: number;
    successRate: number;
    totalRuns: number;
    totalTokens: number;
}
Calculate aggregate statistics for a collection of runs
Parameters
- runs: { duration: number; success: boolean; tokenCount: number }[]
Returns {
    avgDuration: number;
    avgTokens: number;
    successRate: number;
    totalRuns: number;
    totalTokens: number;
}
- Defined in src/lib/audit-analyzer.ts:207

calculateStepDuration

calculateStepDuration(step: MJAIAgentRunStepEntity): number
Calculate step duration from entity timestamps
Parameters
- step: MJAIAgentRunStepEntity
Returns number
- Defined in src/lib/audit-analyzer.ts:33

detectErrorPattern

detectErrorPattern(errorMessages: string[]): string
Detect common error patterns across multiple error messages Finds the longest common substring that appears in all errors
Parameters
- errorMessages: string[]
Returns string
- Defined in src/lib/audit-analyzer.ts:86

estimateCost

estimateCost(tokens: number): number
Estimate cost from token count (using average pricing) Note: This uses an average rate and may not reflect actual costs
Parameters
- tokens: number
Returns number
- Defined in src/lib/audit-analyzer.ts:25

estimateTokenCount

estimateTokenCount(text: string): number
Estimate token count from string (rough approximation) Uses the common rule of ~4 characters per token
Parameters
- text: string
Returns number
- Defined in src/lib/audit-analyzer.ts:16

formatDuration

formatDuration(milliseconds: number): string
Format duration in human-readable format
Parameters
- milliseconds: number
Returns string
- Defined in src/lib/audit-analyzer.ts:188

getTruncationRules

getTruncationRules(detailLevel: string, maxTokens: number): TruncationRules
Get truncation rules based on detail level and max tokens
Parameters
- detailLevel: string
- maxTokens: number
Returns TruncationRules
- Defined in src/lib/audit-analyzer.ts:41

suggestFixes

suggestFixes(
errorPattern: string,
failedSteps: { errorMessage: string; stepType: string }[],
): string[]
Suggest fixes based on error pattern and failed step context Analyzes error messages and provides actionable debugging suggestions
Parameters
- errorPattern: string
- failedSteps: { errorMessage: string; stepType: string }[]
Returns string[]
- Defined in src/lib/audit-analyzer.ts:111

truncateField

truncateField(text: string, maxChars: number): string
Truncate field with smart preview (first + last chars) Shows beginning and end of content with truncation indicator
Parameters
- text: string
- maxChars: number
Returns string
- Defined in src/lib/audit-analyzer.ts:66