Log Tools

Log analysis and processing tools for AI agents, exposed via the Model Context Protocol (MCP).

Tools

parse_logs

Parse log content in common formats: JSON logs, Apache/Nginx access logs, syslog, and structured logs. Supports custom regex patterns with named capture groups. Extracts timestamp, level, message, source, and additional structured fields.

filter_logs

Filter parsed log entries by:

• Log level (ERROR, WARN, INFO, DEBUG)
• Date range (ISO 8601)
• Keyword search (case-insensitive)
• Regex pattern matching
• Source/logger name

Returns matched entries along with level occurrence counts.

analyze_logs

Analyze log patterns to surface operational insights:

• Top 10 most frequent errors (normalized)
• Error rate over time (hourly buckets)
• Request volume per hour
• Slowest endpoints (from HTTP access logs)
• Anomaly detection (sudden spikes or drops using standard deviation)

format_logs

Convert between log formats and produce reports:

• JSON to plain text, plain to JSON, CSV, or table
• ANSI colorization by log level
• Structured field extraction
• Summary report generation

log_stats

Compute comprehensive log statistics:

• Total and non-empty line counts
• Level distribution with percentages
• Date range coverage
• Unique IP addresses and request counts
• Top user agents
• HTTP status code distribution
• Throughput (entries per second)

Setup

npm install
npm run build

Usage with Claude Desktop

Add to your claude_desktop_config.json:

{
  "mcpServers": {
    "log-tools": {
      "command": "node",
      "args": ["path/to/mcp-log-tools/dist/index.js"]
    }
  }
}

Supported Log Formats

Format	Example
JSON	{"timestamp":"2024-01-15T10:30:00Z","level":"ERROR","message":"Connection failed"}
Apache/Nginx	127.0.0.1 - frank [10/Oct/2000:13:55:36 -0700] "GET /index.html HTTP/1.1" 200 2326
Syslog	Jan 1 00:00:00 hostname process[1234]: Something happened
Structured	2024-01-15T10:30:00.000Z [ERROR] [ModuleName] Something failed
Custom	Any format using a regex with named capture groups

License

MIT