Auto-Setup Unit Tests for Your Codebase

Writing unit tests by hand is slow, and most AI test generators stop at *writing* the files and
hand you a pile of red. **This skill writes AND runs them.** Point it at your repo, or at a
specific change; it reads how your project tests, writes idiomatic tests, then runs them through
your own test command and fixes the test-side failures until green.

### ✅ What you get
- **Auto-detected runner + conventions** — Jest/Vitest, pytest, Go `testing`, XCTest, JUnit/Kotest
- **Curated coverage** — it finds the units worth testing and *you* pick which matter (never every file)
- **Idiomatic AAA tests** — happy path · edge cases · error paths · state transitions, in your style
- **Realistic fixtures + in-process mocks** — network/DB/filesystem/clock mocked; deterministic, offline
- **A real run-&-fix-until-green loop** — runs what it wrote, triages, self-fixes test bugs (bounded)
- **A clear report** — files written, pass/fail counts, and exactly which assertion failed when it halts

### 🔁 Writes *and* verifies — not just a scaffolder
Unit tests run in-process with your own runner, so there's nothing to stand up — no mock server,
no device. That means this skill can actually **run the tests it wrote** and iterate to green,
not just drop files and wish you luck.

### 🛡️ Honest by default — it never "fixes" your code to go green
The only files it writes are tests. When a test exposes a **real bug** in the code under test, it
**stops and reports** the exact assertion — what the code did vs. what the test expected — and
lets *you* decide. It will not edit your source or weaken an assertion to force a green run (the
"vibe-coding death spiral" that makes test suites worthless).

### 🧪 Why it works
Distilled from a **production testing harness used on real shipping apps**:
- Matches your existing runner, assertion library, file location, and naming — detected, not imposed
- Asserts **behavior**, not the implementation line-for-line
- Mocks the **boundary**, not the unit under test
- A **bounded** fix loop that knows a test bug (fix it) from an app bug (stop and report)

### ▶️ What you run
Run `/bootstrap-unit-tests`. It detects your runner, scopes the units (your change or a list you
pick), shows the coverage plan, writes nothing until you confirm — then writes the tests and runs
them to green, reporting anything it couldn't resolve.

### 📋 Requirements
- A codebase in a supported ecosystem (JS/TS, Python, Go, Swift, Kotlin/Java)
- Its test runner installed (the skill detects it; it never installs or guesses a command)

---
*Built for developers who want real, passing unit tests stood up in minutes — and who don't want
an AI quietly editing the code to make red turn green.*