AI Agent Streamlines Audit of Automated Tests for a Hospitality Platform
October 30, 2026 · 2 min read
Also published in Português (Brasil)
A testing team utilized an AI agent to conduct an audit of automated tests as they transitioned from Java to .NET. The team, part of a company focused on hospitality automation, faced the challenge of updating approximately 6,500 end-to-end (E2E) tests accumulated over 15 years. The audit aimed to identify tests worth migrating, those requiring fixes, and to eliminate duplicates or obsolete tests before the transition.
The need for an audit was triggered by the planned migration of automated tests to a new technology stack. The team recognized that simply rewriting tests as they were would inherit existing technical debt, prompting the decision to conduct a thorough evaluation of the test suite.
To expedite the audit process, the team integrated an AI agent. The agent was configured within OpenCode to autonomously analyze all necessary data, including the product's source code and run statistics. This setup involved cloning local repositories of the product and the test suite, which allowed the agent to explore the folder structure and identify methods and classes without manual code input.
The integration with TestIT was established through the Model Context Protocol (MCP). Initially, direct API requests were attempted, but the interaction was later shifted to the MCP server, enabling the agent to retrieve necessary entities and metadata from TestIT.
The team evaluated several language models available in OpenCode, ultimately selecting GPT-5.5 for its effectiveness in processing long prompts and accurately utilizing the MCP tools without generating irrelevant information.
After configuring the AI agent, it was tasked with analyzing test scenarios and classifying them into specific categories for migration, integration, or removal. This systematic approach provided clear next steps for the team, improving the efficiency of the audit process.
In the first iteration, the agent was assigned to assess specific tests, categorizing them based on their relevance to the new framework or their functional status. The results were intended to guide the migration strategy, ensuring that only valid and necessary tests would be transitioned to the new .NET framework.