Log Analyzer Skill

Log analysis scans structured JSON logs from pipeline runs to identify errors, warnings, and performance bottlenecks. This Skill parses JSONL files, detects failures, tracks API retries and slow calls, and displays results as readable tables with plain-English summaries so you can quickly understand pipeline health and pinpoint root causes.

Yes. The Skill supports filtering by log level (error, warning, info) and time windows. Use the --level flag to show only specific severity levels and --hours to analyze logs from recent runs, letting you focus on the issues most relevant to your investigation.

The Skill parses structured JSON logs stored as JSONL (JSON Lines) files in the logs/ directory, one JSON object per line. It handles malformed entries gracefully, normalizes timestamps to ET timezone, and reconstructs complete pipeline history from per-script log files.

The Skill automatically detects and highlights API retries and rate-limiting events within your logs, presenting metrics that help you spot performance bottlenecks. This lets you optimize retry strategies and understand whether external services are throttling your pipeline.

The Skill requires the Rich library for tabular display and read-only access to your logs/ directory. No database or complex environment configuration is needed—run the analysis script directly from your pipeline's .claude/skills/ directory.