@tpmjs/tools-slo-draft

Draft Service Level Objective (SLO) definitions for services with comprehensive documentation, error budgets, and alerting strategies.

Installation

npm install @tpmjs/tools-slo-draft

Usage

import { sloDraftTool } from '@tpmjs/tools-slo-draft';

// Use with Vercel AI SDK
const result = await sloDraftTool.execute({
  serviceName: 'Payment API',
  metrics: [
    {
      name: 'Availability',
      target: 99.9,
      window: '30d'
    },
    {
      name: 'API Latency P99',
      target: 95.0,
      window: 'rolling 7 days'
    },
    {
      name: 'Error Rate',
      target: 99.5,
      window: '30d'
    }
  ]
});

console.log(result.slo); // Markdown formatted SLO document
console.log(result.summary); // Brief summary
console.log(result.metrics); // Processed metrics with severity

Input

• serviceName (string, required): Name of the service
• metrics (array, required): Array of metric objects with:
  • name (string): Metric name (e.g., "Availability", "Latency P99")
  • target (number): Target percentage (0-100)
  • window (string): Time window (e.g., "30d", "7 days", "rolling 30 days")

Output

Returns an object with:

• slo (string): Comprehensive SLO document in markdown format
• metrics (array): Processed metrics with additional metadata:
  • Original name, target, window
  • windowType: 'rolling' or 'calendar'
  • severity: 'critical' (≥99.9%), 'high' (≥99%), or 'medium' (<99%)
• summary (string): Brief summary of the SLO
• metadata (object):
  • serviceName: Service name
  • createdAt: ISO timestamp
  • totalMetrics: Total number of metrics
  • criticalMetrics: Number of critical severity metrics

Features

• Automatic Severity Classification: Metrics are classified as critical, high, or medium based on targets
• Error Budget Calculation: Calculates allowed downtime for each metric
• Alerting Strategy: Provides burn rate-based alerting recommendations
• Window Type Detection: Automatically detects rolling vs calendar windows
• Comprehensive Documentation: Generates full SLO document with monitoring and review processes

Example Output

The tool generates a markdown document including:

• SLO definitions with targets and allowed downtime
• Error budget policies
• Alerting strategies based on severity
• Review processes and schedules
• Links to related documentation

License

MIT