Overview

Jtest includes an MCP (Model Context Protocol) Server extension that enables seamless integration between Jtest and AI agents or LLM-based coding assistants (such as GitHub Copilot). This MCP Server can be used within any LLM-enabled development environment that supports MCP integration, including IntelliJ, VS Code, and others. With this extension, an AI agent can interact directly with Jtest, delivering context-aware guidance, analysis, and automation.This extension allows you to use an AI coding assistant to:

Retrieve details about static analysis rules.
Retrieve line coverage information for analyzed code.
Review static analysis violations reported in your project.
Suggest fixes for violations based on provided code snippets or files.
Suggest rules that are appropriate for specific issues.
Access information from the product user guide.

The MCP server launcher is located in the <INSTALL_DIR>/integration/mcp directory.

The Parasoft MCP server uses the stdio transport protocol exclusively. Parasoft MCP tools do not access any external systems or services.

The get_relevant_rules and search_documentation MCP tools require Java 21. See MCP Server Tools Reference for more details.

The following procedures demonstrate how to integrate an LLM-based agent (using Copilot as an example) with Jtest in different environments.

Configuring the Eclipse IDE to Use the MCP Server with Copilot Chat

In Eclipse, go to the Chat window, and select the Agent mode from the combo box.
Click the Configure Tools... button.
In the GitHub Copilot/MCP window, in the MCP Servers field, create a new Server Configuration:

Specify the Server ID.
Specify stdio as the server type.
Set the path to MCP server launcher as the server command: <INSTALL_DIR>/integration/mcp/jtestmcp.bat

Click Save.
The MCP Server tools are added and appear in the tool picker. You can enable or disable the tools you want to use.

Configuring the Intellij IDE to Use the MCP Server with Copilot Chat

In Intellij, go to the Chat window, and select Agent mode from the combo box.
Click the Configure Tools button.
In the Configure Tools window, click + Add More Tools...
A JSON file will be created containing the MCP configuration.
In the JSON File:
1. Specify the Server ID.
2. Specify stdio as the server type.
3. Set the path to MCP server launcher as the server command: <INSTALL_DIR>/integration/mcp/jtestmcp.bat
Click Save.
The MCP Server tools are added and appear in the tool picker. You can enable or disable the tools you want to use.

Configuring the VS Code IDE to Use the MCP Server with Copilot Chat

In VS Code, go to the Command Palette, and choose MCP: Add Server...
Select Command (stdio) and press ENTER to confirm.
Enter the MCP server launcher.
<INSTALL_DIR>/integration/mcp/jtestmcp.bat
Specify the Server ID.
Choose where to save the configuration:
- User Settings - available in all workspaces
- Workspace settings - available in this workspace
A JSON file will be created containing the MCP configuration.
The MCP server entry is added to MCP servers JSON configuration file.

Configuring Copilot to Prefer Jtest MCP Tools in IDEs

When GitHub Copilot is used inside an IDE, such as IntelliJ or VS Code, with a configured Jtest MCP server, Copilot may choose to use Jtest Agent Skills instead of the available Jtest MCP tools. Because Jtest Agent Skills are primarily designed for CLI-based workflows (see Fixing Violations and Improving Test Coverage Using AI Coding Agents), this can result in slower responses and higher resource consumption.

To ensure that Copilot uses Jtest MCP tools when working in an IDE, configure Copilot instructions as described below.

Choose one of the options:
- Create an AGENTS.md file in the root directory of your repository.
- Use Copilot instructions to provide guidance globally or for a specific project:
  - VS Code: Open or create one of the following files:
    - ~/.config/github-copilot/instructions.md
    - .github/copilot-instructions.md.
  - JetBrains IDEs:
    1. Open Settings.
    2. Navigate to Tools > GitHub Copilot > Instructions.

Add the following content to the file/instructions:

Jtest Tool Preference

When working inside an IDE (IntelliJ IDEA, VS Code), always prefer Jtest MCP tools over Jtest Agent Skills to get existing Jtest Static Analysis results, run Static Analysis, or look up rule documentation.

Save the changes.

After the instructions are configured, GitHub Copilot will preferentially use Jtest MCP tools when interacting with Jtest functionality from within the IDE.

Using Instructions in Copilot

In some cases, Copilot does not invoke Parasoft MCP tools. To ensure that Parasoft MCP tools are called, you can add custom instructions.

In the Copilot Chat window, click your profile picture, and click Personal instructions.

Add natural language instructions to the text box, for example:

When analyzing a report containing violations, use the get_violations_from_report_file tool to review the report and provide insights or suggestions based on the findings.
Use the get_rule_documentation tool to get information about the violations, their severity, and possible remediation steps.

With the instructions provided in the example above, Copilot will use the get_violations_from_report_file tool to extract violations from report.xml (rather than attempting to analyze the file on its own) and will use get_rule_documentation to retrieve documentation details.

Click Save.

MCP Server Tools Reference

Tool	Description	Notes
get_relevant_rules	Returns a list of static analysis rules matching the user query. Each entry contains the rule header and a comma-separated list of corresponding rule identifiers.
query_line_coverage	Returns line coverage information for a source file. Analyzes the coverage XML file to provide a list of line numbers based on the query type: 'coverable', 'covered', 'notcovered'.	An absolute path to the coverage XML file must be provided. The server must have local access to the file.
get_rule_documentation	Returns the documentation for the static analysis rule together with a path to the original documentation file for reference.	An exact rule ID must be provided, for example, BD.PB.ARRAY.
get_violations_from_report_file	Extracts violations from a report XML file and outputs them in JSON format along with key details. Results are refined using optional filters: rule ID, severity and source file name.	An absolute path to the report XML file must be provided. The server must have local access to the file.
search_documentation	Searches the documentation to retrieve relevant pages matching the user query. Returns the content of each page and its source URL. To search static analysis rule documentation, use get_rule_documentation or get_relevant_rules tools.