Self-Healing Automated Test System

You are a QA automation engineer. Build a self-healing test system that detects, diagnoses and repairs broken or flaky automated tests with AI. Tech Stack: [INSERT - Python + pytest or TypeScript + Playwright/Jest, FastAPI service] LLM: [INSERT - GPT-4o / Claude 3.5 / local code model] CI: [INSERT - GitHub Actions / GitLab CI / Jenkins] Store: [INSERT - PostgreSQL plus object storage for traces and artifacts] Deliver the following modules in order: 1. Architecture: a CI collector gathers failures and flakiness signals; a diagnosis agent triages; a repair agent proposes patches through pull requests. 2. Data Model: test runs, failure signatures, flake history, root-cause records, healing proposals with status and approval state. 3. Detection APIs: flake detection via rerun statistics, new-failure vs product-bug classification, correlation with the commit/SUT change that broke the test. 4. Healing Engine: selector/locator self-heal with DOM/visual evidence, fixture updates on API contract drift, wait/timing fixes, brittle-assertion rewrites. 5. Governance: every fix shipped as a reviewed PR with evidence, bounded risk scope, no production-code edits, automatic rollback on re-break. 6. Testing & Evals: mutation testing to validate healing quality, staged rollout starting with flaky tests, false-fix rate and recurrence tracking.