Usage Logging

Overview

Session-aware logging infrastructure for tracking operations across plugins. Provides structured JSONL logging with automatic session management for audit trails and analytics.

When to Use

• Need audit trails for operations
• Tracking costs across sessions
• Building usage analytics
• Debugging with operation history

Core Concepts

Session Management

Sessions group related operations:

• Auto-created on first operation
• Timeout after 1 hour of inactivity
• Unique session IDs for tracking

Log Entry Structure

{
  "timestamp": "2025-12-05T10:30:00Z",
  "session_id": "session_1733394600",
  "service": "my-service",
  "operation": "analyze_files",
  "tokens": 5000,
  "success": true,
  "duration_seconds": 2.5,
  "metadata": {}
}

Quick Start

Initialize Logger

from leyline.usage_logger import UsageLogger

logger = UsageLogger(service="my-service")

Log Operations

logger.log_usage(
    operation="analyze_files",
    tokens=5000,
    success=True,
    duration=2.5,
    metadata={"files": 10}
)

Query Usage

# Recent operations
recent = logger.get_recent_operations(hours=24)

# Usage summary
summary = logger.get_usage_summary(days=7)
print(f"Total tokens: {summary['total_tokens']}")
print(f"Total cost: ${summary['estimated_cost']:.2f}")

# Recent errors
errors = logger.get_recent_errors(count=10)

Integration Pattern

# In your skill's frontmatter
dependencies: [leyline:usage-logging]

Standard integration flow:

1. Initialize logger for your service
2. Log operations after completion
3. Query for analytics and debugging

Log Storage

Default location: ~/.claude/leyline/usage/{service}.jsonl

# View recent logs
tail -20 ~/.claude/leyline/usage/my-service.jsonl | jq .

# Query by date
grep "2025-12-05" ~/.claude/leyline/usage/my-service.jsonl

Detailed Resources

• Session Patterns: See modules/session-patterns.md for session management
• Log Formats: See modules/log-formats.md for structured formats

Exit Criteria

• Operation logged with all required fields
• Session tracked for grouping
• Logs queryable for analytics