The Autofix plugin leverages LLM tools to fix code violations in your Java projects running within automated CI/CD pipelines, based on a generated analysis report. Jtest identifies violations, uses LLM integration to propose fixes for those violations, and verifies each proposed fix to ensure that all issues are fully resolved and no new problems are introduced. Successful fixes result in automatically created commits in the working branch of your project source control repository, enabling manual review and approval before merging.

The Jtest Autofix Python script is located in <INSTALL_DIR>/integration/aider.

Prerequisites

The following are required to use this functionality:

Python 3.10-3.12
A version control system
Your secret LLM provider API key

This feature depends on LLMs, which may produce inaccurate information

Jtest Autofix Configuration

Preparing a Virtual Environment

We strongly recommend running the JtestAutoFix.py script inside a Python virtual environment to isolate it from your system-wide Python installation and prevent dependency conflicts.

Follow these steps to set up your virtual environment:

Windows

Create a virtual environment:
```
python -m venv venv
```
Enter the virtual environment:
```
CALL venv\Scripts\activate.bat
```

Install the required dependencies:

python -m pip install --upgrade pip
pip install -r requirements.txt

Linux/Unix

Create a virtual environment:
```
python -m venv venv
```
Enter the virtual environment:
```
source venv/bin/activate
```

Install required dependencies:

python -m pip install --upgrade pip
pip install -r requirements.txt

Autofix Configuration Options

The JtestAutoFix.py script can be configured using environment variables or command-line arguments. Command-line arguments take precedence over environment variables when both are specified.

Below is the complete reference for all available configuration options:

Option	Required	CLI Argument	Environment Variable	Default Value	Description
LLM API Key	Yes	N/A	`%PROVIDER%_API_KEY=<KEY>`	-	Your secret LLM provider API key required by Aider. Replace `%PROVIDER%` with the provider of your choice. Learn more about providers and variables.
Data file	Yes	`--data DATA_FILE_PATH`	N/A	-	Path to the Jtest `.data.json` file. It must match the project used for the baseline analysis report.
Baseline Report	Yes	`--report XML_REPORT_PATH`	`REPORT=<XML_REPORT_PATH>`	-	Path to the Jtest XML report containing violations to be fixed.
Tool Home	No	`--tool-home JTEST_HOME`	`TOOL_HOME=<JTEST_HOME>`	Jtest directory	Path to the Jtest installation directory. By default, it is set to the "../.." directory where the script is located.
Fix Limit	No	`--fix-limit COUNT`	`FIX_LIMIT=<COUNT>`	0 (no limit)	Maximum number of violation fix attempts per run.
Max Attempts	No	`--max-attempts MAX_ATTEMPTS`	N/A	3	Maximum retry attempts per violation before stopping.
Model Name	No	N/A	`MODEL_NAME=MODEL`	gpt-4o-mini	AI model for fixing violations (e.g., gpt-4, gpt-4o, claude-3-sonnet). Use `aider --list-models <fragment>` to see the available models.
Weak Model Name	No	N/A	`WEAK_MODEL_NAME=MODEL`	-	Lightweight model for auxiliary tasks, such as commit message generation. When not specified, it defaults to the specified model name (`MODEL_NAME`).
Verification Command	No	`--verification-command COMMAND`	N/A	-	Command to verify code integrity after each fix (e.g., `"mvn compile"` or `"./gradlew assemble"`). Success is determined by exit code (0 = success).
No Source Control	No	`--allow-no-source-control`	N/A	Disabled	Bypass source control warnings. Use with caution - changes will be made without version control and cannot be reverted.
Help	No	`-h` or `--help`	N/A	N/A	Display help information for the `JtestAutoFix.py` script.

Configuring the Environment

You can use the provided .env.template file as a starting point to create your own .env file with the necessary environment variables.

Running Autofix

Before running the Autofix script, we recommend creating a dedicated branch for the automated fixes. This allows you to review and test all changes before merging them into your main branch.

Basic Usage for a Gradle project

Windows

Set your API key:

SET OPENAI_API_KEY=your_secret_open_ai_key

Execute Autofix with the recommended settings:

python JtestAutoFix.py
 --report "<PROJECT_PATH>/build/jtest/report.xml"
 --max-attempts 3
 --data "<PROJECT_PATH>/build/jtest/jtest.data.json"
 --tool-home "C:\Program Files\Parasoft\Jtest\2025.2"

Linux/Unix

Set your API key:

export OPENAI_API_KEY=your_secret_open_ai_key

Execute Autofix with the recommended settings:

python3 JtestAutoFix.py
 --report "<PROJECT_PATH>/build/jtest/report.xml"
 --max-attempts 3
 --data "<PROJECT_PATH>/build/jtest/jtest.data.json"
 --tool-home "/opt/parasoft/jtest/2025.2"

Advanced Usage Examples

Windows with the Verification Command

Set the environment variables:

SET OPENAI_API_KEY=your_secret_open_ai_key
SET MODEL_NAME=gpt-4.1
SET WEAK_MODEL_NAME=gpt-4o-mini

Execute Autofix with build verification:

python JtestAutoFix.py
 --report "C:\projects\myapp\build\jtest\report.xml"
 --data "C:\projects\myapp\build\jtest\jtest.data.json"
 --tool-home "C:\Program Files\Parasoft\Jtest\2025.2"
 --verification-command "gradle build"
 --fix-limit 10
 --max-attempts 2

Linux/Unix with the Verification Command

Set the environment variables:

export OPENAI_API_KEY=your_secret_open_ai_key
export MODEL_NAME=gpt-4.1
export WEAK_MODEL_NAME=gpt-4o-mini

Execute Autofix with build verification:

python JtestAutoFix.py
 --report "/home/user/projects/myapp/build/jtest/report.xml"
 --data "/home/user/projects/myapp/build/jtest/jtest.data.json"
 --tool-home "/opt/parasoft/jtest/2025.2"
 --verification-command "./gradlew build"
 --fix-limit 10
 --max-attempts 2

To ensure accurate results and efficient use of the Autofix plugin, follow these best practices:

Generate a new baseline after each Autofix session: Code changes made during the autofix process can alter file line numbers and structure, making the violation locations in your original report.xml obsolete. Regenerate your baseline report to ensure it accurately reflects your updated codebase.

Fix history tracking: The script creates data files in the .cliAutoFix folder to track fix attempts and outcomes in *.history.yaml files.

Preserve history: To avoid re-attempting the same violations:

Preserve the .cliAutoFix folder and its *.history.yaml files between runs.
Alternatively, ensure all attempted violations are either fixed or suppressed before the next run.

Reviewing Results

After each successful fix, the Autofix script creates a commit in your repository. Since AI-generated code may contain errors or suboptimal solutions, it is crucial to review all changes carefully.

Follow these best practices when reviewing AI-generated fixes:

Read the commit messages - Understand what changes were made and why.
Test the changes - Ensure the fix resolves the violation without introducing new issues.
Check the code quality - Verify that the fix follows your team's coding standards.
Apply incremental fixes - For minor issues in otherwise good fixes, create separate commits with manual corrections.
Revert when necessary - If a fix is fundamentally wrong, revert the entire commit.
Use a branching strategy - Work on a feature branch and merge after review, or cherry-pick verified commits.

Troubleshooting

If you encounter any issues while running the script, ensure that:

All dependencies are correctly installed
Environment variables are properly configured
All required CLI parameters are properly passed
Linux/Unix only: Ensure that Python 3.10+ is installed and accessible via the python3 or python commands.
Linux/Unix only: Check the file permissions for the script and data files.

Page tree

Fixing Violations Using Autofix