I need an AI-driven assistant, written in Python, that sits on top of a Git repository and reacts to every new commit. Each time code changes land, the agent should: • Analyse the diff, understand what was modified or added, and automatically draft the corresponding manual test cases as well as Selenium functional scripts. When a matching script already exists, it must smart-merge or refactor it rather than creating a duplicate. • Do the same for performance testing: generate fresh k6 scripts and LoadRunner script, or update the existing ones so they stay in sync with the latest code paths. • Keep a lightweight inventory (YAML or JSON) of every test artifact it creates or updates for future traceability. • Push the new or amended scripts back to the repo through a standard Git workflow (pull request or direct commit, configurable). Acceptance criteria 1. Point the agent at a sample Git project; after a commit it must open a pull request containing at least one generated Selenium file and one k6 script, both runnable without manual edits. 2. On a second commit touching the same code, the agent updates—not duplicates—the previous scripts. 3. Clear console or log output showing reasoning steps, and a README that explains how to redeploy the tool in another repo.