Log Analyzer

A CLI tool for analyzing and comparing JSON logs from the Applitools testing framework.

Installation

# Auto-detect platform and install latest release
curl -fsSL https://raw.githubusercontent.com/eirenik0/log-analyzer/main/scripts/install.sh | bash

# Or build from source
cargo install --path .

Quick Start

# Compare two log files
log-analyzer compare file1.log file2.log

# Show only differences
log-analyzer diff file1.log file2.log

# Get log file overview
log-analyzer info file.log

# Analyze performance bottlenecks
log-analyzer perf file.log

# Generate LLM-friendly output
log-analyzer llm file.log

Commands

Command	Aliases	Description
compare	cmp	Compare two log files
diff		Compare showing only differences
info	i, inspect	Display log file statistics
perf		Analyze operation timing
process	llm	Generate LLM-friendly JSON output
llm-diff		Generate LLM-friendly diff output

Global Options

Option	Env Variable	Description
-F, --format <text|json>	FORMAT	Output format
-j, --json	JSON	JSON output (shorthand for -F json -c)
-c, --compact	COMPACT	Compact output mode
-f, --filter <expr>	FILTER	Filter expression (see below)
-o, --output <path>		Output file path
--color <auto|always|never>		Color output control
-v, --verbose		Increase verbosity
-q, --quiet		Show only errors

Filter Expression Syntax

Use -f, --filter with a unified expression syntax:

--filter "type:value [!type:value] ..."

Filter types (with aliases):

Type	Aliases	Description
component	comp, c	Filter by component name
level	lvl, l	Filter by log level (INFO, ERROR, etc.)
text	t	Filter by text in message
direction	dir, d	Filter by direction (incoming/outgoing)

Prefix with ! to exclude. Multiple filters use AND logic.

# Only core-universal component
-f "c:core-universal"

# Only ERROR level logs
-f "l:ERROR"

# Core component, exclude DEBUG level
-f "c:core !l:DEBUG"

# Contains 'timeout', incoming direction only
-f "t:timeout d:incoming"

Command-Specific Options

compare / diff

Option	Description
-D, --diff-only	Show only differences (always on for diff)
--full	Show full JSON objects
-s, --sort-by	Sort by: time, component, level, type, diff-count

info

Option	Description
-s, --samples	Show sample messages per component
--json-schema	Display JSON schema information
-p, --payloads	Show payload statistics
-t, --timeline	Show timeline analysis

perf

Option	Description
--threshold-ms <ms>	Slow operation threshold (default: 1000)
--top-n <number>	Number of slowest operations (default: 20)
--orphans-only	Show only unfinished operations
--op-type <Request|Event|Command>	Filter by operation type

Sort options: duration, count, name

llm / llm-diff

Option	Description
--limit <number>	Max entries (default: 100, 0 = unlimited)
--no-sanitize	Disable sensitive field hiding

Examples

# Compare logs filtering by component and level
log-analyzer diff file1.log file2.log -f "c:core-universal l:ERROR"

# Exclude DEBUG logs from comparison
log-analyzer diff file1.log file2.log -f "!l:DEBUG"

# Save JSON diff to file
log-analyzer -j -o diff.json diff file1.log file2.log

# Show operations slower than 500ms
log-analyzer perf file.log --threshold-ms 500

# Comprehensive log analysis
log-analyzer info file.log --samples --timeline --payloads

# LLM-friendly diff with custom limit
log-analyzer llm-diff file1.log file2.log --limit 50

Environment Configuration

Set defaults via environment variables (prefix LOG_ANALYZER_):

export LOG_ANALYZER_FORMAT=json
export LOG_ANALYZER_FILTER="!l:DEBUG"
export LOG_ANALYZER_COMPACT=true

Claude Code Integration

Installation

Install the Claude Code skill to use interactive log analysis in any project:

/plugin marketplace add https://github.com/eirenik0/log-analyzer
/plugin install log-analyzer

Usage

Use the /analyze-logs command in Claude Code for interactive analysis:

/analyze-logs diff file1.log file2.log          # Compare and explain differences
/analyze-logs perf test.log --threshold-ms 500  # Find performance bottlenecks
/analyze-logs info test.log --samples           # Log structure overview
/analyze-logs llm test.log                      # Generate LLM-friendly output

Features

• Structured parsing - Extracts and parses JSON payloads automatically
• Semantic comparison - Compares JSON objects regardless of property order
• Advanced filtering - Include/exclude by component, level, content, or direction
• Performance analysis - Identify slow and orphan operations
• LLM-friendly output - Sanitized, compact JSON for AI consumption
• Flexible output - Text or JSON format with color and verbosity control