test-looplisted

# Test Loop — Write · Run · Fix Until Green ## Purpose Two related workflows around automated tests, both ending in a **green run**: - **Coverage mode** — write tests for existing code, run, and fix failures. - **TDD mode** — write the tests *first* from the spec, then implement the code until every test passes. **Language:** explanations in Hebrew; test code/comments in English. --- ## Step 0 — Detect the test setup (always first) - Identify the framework from the project: `package.json` scripts + devDeps (Jest, Vitest, Playwright, Mocha), `pytest`/`unittest` (Python), `go test`, `cargo test`, RSpec, etc. **Never assume** — read the config. - Find the run command (`npm test`, `pytest -q`, `go test ./...`) and the test file convention (`*.test.ts`, `*.spec.ts`, `test_*.py`, `_test.go`) and location. - If no test framework exists and the task needs one, propose + install the ecosystem-standard choice before writing tests, then scaffold minimal config. - Read an existing test file to **match style, imports, and patterns** exactly. --- ## Coverage Mode — tests for existing code Trigger shape: "write tests for `app/parsers/feed.py`, run them, and fix any failures". 1. **Read the target thoroughly** — understand its real behavior, inputs, outputs, error paths, and dependencies (what to mock vs. call for real). 2. **Write meaningful tests**, not filler: - Happy path(s) with representative inputs. - Edge cases: null/undefined, empty, zero, negative, boundar