🔍

io.github.Fato07/log-analyzer-mcp

AI-powered log analysis - parse, search, extract errors across 9+ formats

0 installs

Trust: 37 — Low

Ask AI about io.github.Fato07/log-analyzer-mcp

I know everything about io.github.Fato07/log-analyzer-mcp. Ask me about installation, configuration, usage, or troubleshooting.

0/500

Loading tools...

Reviews

Documentation

Log Analyzer MCP

🔍 Stop copy-pasting logs into AI. Let Claude read them directly.

An MCP server for AI-powered log analysis. Parse, search, and debug log files across 9+ formats — right from Claude Code.

📊 At a Glance


14 MCP tools	9+ log formats
280 tests	81%+ coverage

🎬 Demo

Log Analyzer MCP Demo

Analyzing logs with 14 specialized tools

🤔 Why?

Without log-analyzer-mcp	With log-analyzer-mcp
Copy-paste chunks of logs	Point Claude at the file
Lose context between pastes	Full file access
Manual format parsing	Auto-detection
Miss related errors	Smart correlation

✨ Features

Auto-Detection — Identifies format from 9+ common log types
Smart Search — Pattern matching with context, regex, and time filtering
Error Extraction — Groups similar errors, captures stack traces
Natural Language — Ask questions like "what errors happened today?"
Sensitive Data Scan — Detect PII, credentials, and secrets
Multi-File Analysis — Correlate events across distributed systems
Streaming — Handles 1GB+ files without memory issues

🚀 Quick Start

# Install (adds to Claude Code automatically)
uvx codesdevs-log-analyzer install

Then in Claude Code:

Analyze /var/log/app.log and tell me what's causing the errors

📦 Installation

One-liner (Recommended)

uvx codesdevs-log-analyzer install

Manual

pip / uv / Claude Code config

# pip
pip install codesdevs-log-analyzer

# uv
uv tool install codesdevs-log-analyzer

Add to ~/.claude/settings.json:

{
  "mcpServers": {
    "log-analyzer": {
      "command": "uvx",
      "args": ["codesdevs-log-analyzer"]
    }
  }
}

📋 Supported Formats

Format	Example
Syslog	`Jan 15 10:30:00 hostname process[pid]: message`
Apache/Nginx	`127.0.0.1 - - [15/Jan/2026:10:30:00] "GET /path" 200`
JSON Lines	`{"timestamp": "...", "level": "ERROR", "message": "..."}`
Docker	`2026-01-15T10:30:00.123Z stdout message`
Python	`2026-01-15 10:30:00,123 - module - ERROR - message`
Java/Log4j	`2026-01-15 10:30:00,123 ERROR [thread] class - message`
Kubernetes	`level=error msg="..." ts=2026-01-15T10:30:00Z`
Generic	Any line with recognizable timestamp

⚡ Performance

Metric	Value
100MB log file	< 10 seconds
Memory footprint	Streaming (no full load)
Max tested size	1GB+
Format detection	< 100ms

🛠️ Available Tools

Tool	Description
`log_analyzer_parse`	Detect format and extract metadata
`log_analyzer_search`	Search with context lines
`log_analyzer_extract_errors`	Extract and group errors
`log_analyzer_summarize`	Generate debugging summary
`log_analyzer_correlate`	Find related events
`log_analyzer_watch`	Monitor for new entries
`log_analyzer_ask`	Natural language queries
`log_analyzer_scan_sensitive`	Detect PII/credentials
+ 6 more	Full reference →

💡 Examples

Find errors:

Extract all errors from /var/log/app.log, group similar ones

Search with context:

Search for "timeout" in app.log with 5 lines of context

Correlate events:

What happened 60 seconds before each OutOfMemoryError?

Scan for secrets:

Check /var/log/app.log for accidentally logged credentials

🔧 Development

git clone https://github.com/Fato07/log-analyzer-mcp
cd log-analyzer-mcp
uv sync
uv run pytest -v --cov

📈 Star History

📄 License

MIT License - see LICENSE for details.

Found this useful? Give it a ⭐ on GitHub!

Report bugs · Request features · Discussions · Full docs

Built by Fato07 at CodesDevs