Do you want to automate generating the manual test cases using LLMs?

Author: Payani Putturu, Senior Manager – QA at Hashagile Technologies

In the era of agile delivery and continuous testing, every minute spent on repetitive documentation is a minute lost from actual quality assurance. One of the most time-consuming QA tasks is Creating manual test cases from Acceptance Criteria.

Recently, we implemented a Proof of Concept (POC) that automated this process using:

• JIRA MCP Server (Model Context Protocol)

• Cline (Visual Studio Code extension)

• Large Language Models (LLMs)

The result? Test cases were generated in seconds — directly from a JIRA ticket — and added back to the ticket, following our custom template.

🎯Objectives

The POC aimed to:

• Connect to JIRA securely via MCP Server using Cline

• Retrieve ticket details (especially Acceptance Criteria)

• Automatically generate structured test cases using a predefined template

• Post the generated cases back to JIRA under a designated field or comment section

📜Prerequisites

• Access to a JIRA instance (Cloud or Server) from Acceptance Criteria.

• MCP Server (Model Context Protocol)

• Cline Visual Studio Code extension

• API Token or basic authentication for JIRA

• A Test Case Template in a table format.

⚙️Tools at a Glance

Key Technologies Overview:

• MCP Server – A standardized middleware that creates secure bridges between enterprise tools (like JIRA) and AI workflows. MCP servers act as translators, converting AI requests into proper API calls while maintaining security and access controls. Think of MCP as a universal translator that allows AI assistants to communicate with your business tools safely.

• Cline Extension – A sophisticated Visual Studio Code extension that enables AI-powered automation within your development environment. Cline can execute commands, read files, and interact with external systems through MCP servers, making it possible to create complex automation workflows directly from your IDE.

📈Step-by-Step Implementation

Step 1: Connect Cline to MCP Server

• Install the Cline Extension from the VS CODE Extensions Store.

• Click on the Cline icon of left pane of the VS Code IDE.

• Set up the MCP Server and configure it with your JIRA instance credentials.

• In Cline, establish a secure connection to the MCP Server using the server’s URL and authentication.

You can refer the below steps for configuration

• Click on “Manage MCP Severs”

• Click on Settings icon 

• Click on “Configure MCP Servers” which opens cline_mcp_settings.json file. The following is the JIRA MCP Server sample config snippet. 

{
  "mcpServers": {
    "mcp-atlassian": {
      "command": "uvx",
      "args": [
        "mcp-atlassian",
        "--jira-url=https://youOrgName.atlassian.net",
        "--jira-username=YourName@example.com",
        "--jira-token=YouJiraToken"
      ],
      "autoApprove": ["jira_get_all_projects"]
    }
  }
}

• Please visit https: // id.atlassian.com/manage-profile/security/api-tokens to generate Jira Access token.

• Please visit https: // docs .astral.sh/uv/getting-started/installation/#pypi to install uvx

Step 2: Configure API Provider

• Configure the API Provider based on your convenience.

     Note: We have tried with pro version keys. You can refer the following for choosing AI Model based on your preference

Step 3: Retrieve Ticket Details

• In Cline, use the available commands or scripting panel to fetch a JIRA ticket using its key (e.g., CLIN-1).

• Sample Prompt:

“Fetch the description of CLIN-1 ticket”

• AI understands your request based on your prompt and then it sends the request with the arguments as below and it starts processing the same.
• {“issue_key”: “CLIN-1”, “fields”: “description”}

• Approve/Reject the request.
  • On Approve, Cline starts fetching the ticket description.
  • On Reject, Cline rejects the request.

Step 4: Generate Test Cases and Push back to JIRA ticket

• In Cline, use the available commands or scripting panel to prompt to generate the test cases as per the template you define and add the generated test cases to JIRA ticket.

• Sample Prompt:

         “Generate the Functional test cases and add the same in the CLIN-1 under Comments section as per the below template in a table format

Test Case ID, Pre-Condition, Test Case Title, Steps, Expected Results”

• Go to JIRA ticket and refer the comments section for generated test cases.

Note:

• You can even tell Cline to assign the ticket to team member through prompt.

• You can prompt to write the Non-Functional test cases like Performance and Security (Penetration) tests as well. Based on the Acceptance Criteria of the ticket, test cases will be generated.

Important Note:

• Having clearly defined Acceptance Criteria would give you better test cases.

🏆Benefits

• ⏱Time Savings – Reduced case creation time from ~30 mins to under 2 mins per ticket

• 📈 Improved Coverage – AI ensures no acceptance criteria are overlooked

• ⚡ Consistency – Standardized format for all manual test cases 

• 🔄 Extendability – Can be adapted for automated test script generation in the future 

💡Conclusion 

This POC shows how MCP Server + Cline + LLMs can eliminate a large chunk of repetitive QA work.

By embedding AI in our QA workflow, we not only speed up documentation but also improve consistency and coverage.