Excessive unit test writing creates redundant code and slows development

Build a React‑based SPA that authenticates via OAuth2 to GitHub/GitLab and displays a per‑file test matrix. The backend is a Python FastAPI service running in a Docker container. For each supported language (currently JavaScript/TypeScript and Python) use the Babel parser (via @babel/parser) and ast module respectively to generate an AST for production files and test files. Instrument the code with Istanbul/nyc (JS) and coverage.py (Python) to collect branch‑level coverage data on a nightly CI run, storing the resulting JSON reports in a DuckDB instance for fast relational queries. A custom overlap engine reads the coverage tables, groups tests by identical set of covered branches, and computes a relevance score using a weighted formula: unique branches * 0.7 + recent failure count * 0.3 (failure data pulled from the CI’s test result API). The FastAPI endpoint `/overlap` returns JSON with test IDs, overlap groups, and scores. A GitHub App webhook triggers on `pull_request` events; the service runs a diff‑only coverage analysis using `nyc instrument` on the PR’s changed files, compares against existing coverage, and posts inline comments via the GitHub REST API (`POST /repos/{owner}/{repo}/pulls/{pull_number}/comments`). Authentication uses GitHub App JWTs. The React UI consumes `/overlap` and `/pr-feedback` endpoints, rendering a heat‑map matrix with D3.js and allowing users to mark tests for deletion, which calls a FastAPI `/prune` endpoint that opens a new branch with a PR that removes the selected test files.

By automatically surfacing duplicate test cases and scoring test relevance, the tool cuts the time engineers spend on test maintenance, allowing faster iteration on feature work. The overlap detection reduces CI runtime because fewer redundant tests run, lowering compute costs. Inline PR feedback prevents redundancy from entering the codebase, preserving test suite health and reducing human error in test design, ultimately leading to a leaner, more reliable test ecosystem.