Error Analyzer MCP Server 🔧

A specialized Model Context Protocol (MCP) server that analyzes errors in your projects, identifies root causes, and generates detailed fix plans with a built-in approval workflow before applying any changes.

🌟 Overview

When your project encounters errors, this MCP server helps you:

• Read and analyze log files to understand what went wrong
• Identify the affected files and their dependencies
• Diagnose the root cause of errors
• Create detailed fix plans with step-by-step solutions
• Request your approval before making ANY file modifications
• Apply approved fixes safely to your project

This server emphasizes safety and transparency: it never modifies your files without your explicit approval and clear understanding of what will change.

🏗️ Architecture

• Reads log files from a configurable logger directory
• Analyzes error messages and stack traces
• Identifies related files and dependencies
• Generates fix plans with risk assessment
• Implements a two-phase workflow:
  1. Analysis & Planning Phase: Creates fix plans
  2. Approval Phase: Waits for your explicit approval
  3. Application Phase: Applies approved changes

🛠️ Available Tools

1. list_available_logs

List all log files in your logger directory.

• Returns all .log files available for analysis

2. analyze_error

Analyze errors in a specific log file.

• Input: project directory path, log file name
• Returns: error location, root cause, affected files, severity level
• Creates an analysis ID for further processing

3. create_fix_plan

Create a detailed work plan to fix identified errors.

• Input: analysis ID from previous step
• Returns: fix plan with proposed fixes, impact assessment, risk level
• Status: Awaiting your approval
• Creates a plan ID for tracking

4. approve_fix_plan

Approve a fix plan after reviewing the proposed changes.

• Input: plan ID
• Effect: Marks the plan as approved
• Next step: Call apply_fix_plan to make the changes

5. reject_fix_plan

Reject a fix plan if you disagree with proposed solutions.

• Input: plan ID, optional feedback
• Effect: Cancels the plan
• Next step: Can request new analysis with different parameters

6. apply_fix_plan

Apply the approved fixes to your project files.

• Input: plan ID (must be approved first)
• Effect: MODIFIES YOUR PROJECT FILES
• Returns: List of changes applied
• Safety: Only works if plan was explicitly approved

7. get_fix_plan_details

Retrieve detailed information about a specific fix plan.

• Useful for reviewing plans before approval

8. list_pending_plans

See all fix plans created in this session with their statuses.

• Helps track pending approvals and applied fixes

📋 Workflow Example

1. Error occurs in your project
   ↓
2. Error is written to logger/error.log
   ↓
3. Call analyze_error(project_dir="/path/to/project", log_file_name="error.log")
   → Returns: analysis_id = "abc123"
   ↓
4. Call create_fix_plan(analysis_id="abc123")
   → Returns: plan_id = "xyz789"
   → Status: "AWAITING USER APPROVAL"
   ↓
5. Review the proposed fixes in the plan
   ↓
6. Either:
   a) Call approve_fix_plan(plan_id="xyz789") → then apply_fix_plan(plan_id="xyz789")
   b) Call reject_fix_plan(plan_id="xyz789", feedback="...")

🚀 Setup

Installation

1. Clone or extract the project to your workspace:

cd m:\mcp\mcp-read-log

2. Install dependencies:

pip install -r requirements.txt

3. Configure environment (optional):

# Copy the example file
cp .env.example .env

# Edit .env to customize:
# - LOG_DIR: Path to your logger directory (default: ./logger)
# - PROJECT_DIR: Default project to analyze
# - BACKUP_BEFORE_APPLY: Create backups before applying fixes

4. Create logger directory if it doesn't exist:

mkdir logger

📖 Usage with Claude or Other AI Agents

In Claude or Cline:

You have access to the error-analyzer-mcp MCP server. When a user's project has errors:

1. First call: list_available_logs() to see what log files are available
2. Then call: analyze_error() with the project path and log file name
3. Review the analysis results
4. Call: create_fix_plan() with the analysis ID
5. Show the user the proposed fixes and ask for approval
6. Only after explicit approval, call: apply_fix_plan()

Example Session:

User: "My Python project is broken with errors"
AI: Calling list_available_logs()... Found: error.log, debug.log
AI: Which log should I analyze? Let me check error.log.
AI: Calling analyze_error(project_dir=".", log_file_name="error.log")
AI: Analysis complete! Found SyntaxError on line 42, affects 3 files.
AI: Calling create_fix_plan()...
AI: Created fix plan. Here are the proposed changes:
   - Fix syntax error in main.py line 42
   - Update import in handler.py line 15
AI: This has LOW risk. Do you approve these changes?
User: "Yes, apply them"
AI: Calling approve_fix_plan()... then apply_fix_plan()
AI: Done! Applied 2 fixes to your project.

🔒 Safety Features

✅ No file modifications without approval

• Every fix plan requires explicit user approval before changes

✅ Clear risk assessment

• Each plan shows risk level, severity, and dependencies

✅ Transparent change tracking

• See exactly which files and lines will be modified

✅ Reversible operations

• Backup support (configurable in .env)
• Can review changes before committing

📝 Log File Format

Place log files in the logger directory. Supported formats:

• Standard error logs with file:line:column notation
• Stack traces with file paths
• IDE error output
• Custom formatted logs with error messages

Example:

Error in /path/to/file.py on line 42:
  SyntaxError: invalid syntax
    x = y +
          ^
File "helper.py", line 15, in process()
  undefined_variable = 5

🔧 Configuration Files

pyproject.toml

Project metadata and Python dependencies. Uses PyPI packages for MCP, Pydantic, and python-dotenv.

requirements.txt

Direct pip requirements for easy installation.

settings.py

Configuration dataclass for runtime settings.

.env

Environment variables:

• LOG_DIR: Custom logger directory path
• PROJECT_DIR: Default project to analyze
• BACKUP_BEFORE_APPLY: Whether to create backups (true/false)

📂 Project Structure

mcp-read-log/
├── main.py                          # MCP server with all tools
├── settings.py                      # Configuration management
├── pyproject.toml                   # Project metadata
├── requirements.txt                 # Python dependencies
├── .env.example                     # Example configuration
├── README.md                        # This file
│
├── models/
│   ├── result.py                   # ToolResult, ErrorInfo, ErrorAnalysis models
│   └── error_analysis_models.py    # Input models for MCP tools
│
├── services/
│   ├── log_reader.py               # Log file reading utilities
│   └── error_analysis_service.py   # Core error analysis logic
│
├── utils/
│   └── (utilities for path, validation, errors)
│
├── tests/
│   └── (unit tests for services)
│
└── logger/                          # Log files (created on first use)
    ├── error.log
    └── debug.log

🐛 Debugging

If the MCP server doesn't start:

1. Check Python version: Requires Python 3.12+

python --version

2. Verify dependencies:

pip list | grep mcp
pip list | grep pydantic

3. Test log reading:

python -c "from services.log_reader import list_log_files; print(list_log_files())"

4. Check environment:

echo %LOG_DIR%  # Windows
echo $LOG_DIR   # Linux/Mac

📄 License

Based on the git-mcp-server framework by the course instructor.

🤝 Contributing

To extend this MCP server:

• Add new error detection patterns in error_analysis_service.py
• Add new analysis types by extending ErrorAnalysis model
• Add new tool methods in main.py following the @mcp.tool() pattern

---

Remember: This server helps you understand and fix errors safely. Always review proposed changes before approval!