Class ReviewFeedbackTool

constructor

• new ReviewFeedbackTool(): ReviewFeedbackTool

  Returns ReviewFeedbackTool

annotations

description

name

title

execute

• execute(args: Record<string, unknown>): Promise<ReviewFeedbackOutput>

  Execute review-feedback tool to analyze collected transformation feedback

  This method loads feedback sessions from configured storage locations (project, platform, or both), filters and aggregates the feedback, detects patterns (success/failure/quality), excludes outliers, and returns actionable insights for AI learning and improvement.

  Parameters

  • args: Record<string, unknown>

    Tool input arguments conforming to ReviewFeedbackInput interface

    • sessionId

      Optional specific session ID to analyze (default: all sessions)

    • category

      Optional filter by rule category (e.g., 'standalone', 'solid')

    • since

      Optional ISO 8601 date filter for recent feedback

    • analysisType

      Type of analysis: 'success-patterns', 'failure-patterns', 'quality-trends', or 'all'

    • minConfidence

      Minimum confidence threshold 0.0-1.0 (default: 0.7)

    • minCompleteness

      Minimum completeness threshold 0-100 (default: 70)

    • source

      Storage location strategy: 'project', 'platform', or 'both' (default: 'both')

  Returns Promise<ReviewFeedbackOutput>

  ReviewFeedbackOutput containing:

  • sessions: Number of sessions analyzed
  • transformations: Number of valid transformations analyzed (after filtering)
  • successRate: Overall success rate (0.0-1.0)
  • sourcesRead: Array of directory paths where sessions were read from
  • insights: Array of FeedbackInsight objects with detected patterns sorted by confidence
  • qualityWarnings: Array of QualityWarning objects for data quality issues

  Throws

  Error if feedback storage initialization fails or critical errors occur

  Example

  // Analyze all sessions from both locations
  const result = await tool.execute({
    analysisType: 'all',
    minConfidence: 0.8,
    minCompleteness: 80
  });
  
  // Analyze recent standalone migrations
  const result = await tool.execute({
    analysisType: 'success-patterns',
    category: 'standalone',
    since: '2025-12-01T00:00:00Z',
    source: 'project'
  });
  Copy

  Remarks

  • Automatically excludes outlier feedback detected by OutlierDetector
  • Deduplicates sessions when reading from multiple locations
  • Pattern detection thresholds: success (≥3 occurrences, ≥80% success), failure (≥2 occurrences)
  • The platform store is read only when ANGULAR_MODERNIZER_PLATFORM_ROOT is set
  • Insights are sorted by confidence in descending order