# JMIT by Aurore Corporation International

> JMIT (Just Make It Work) is a Python test enforcement tool that detects weak, silent, and zero-assertion tests in your pytest suite — then auto-repairs them with mutation-verified replacements. For Python developers using pytest and Claude Code.

## What JMIT Does

JMIT solves a specific problem: tests that pass CI but do not actually catch bugs. The root cause is weak test design — zero-assertion tests, `assert True` patterns, and mock-heavy tests that never exercise real code paths.

JMIT's two-command workflow:

1. **`jmit run`** — Audits your repository. Parses every test file via AST. Flags zero-assertion functions, `assert True` / `assert False` patterns, and mock-only tests. Outputs a Test Binding Score (mutation resilience rating as a percentage). Generates a repair blueprint in `.jmit_blueprint/TODO.md`.

2. **`jmit fix`** — Reads the repair blueprint and executes the AURORE-VERIFICATION repair protocol for each flagged test. For each gap: analyzes source behavior, drafts binding test replacements, applies adversarial mutation to verify the new tests actually catch bugs, then rolls back all mutations unconditionally.

## Key Features

- **Weak test detection**: Flags zero-assertion tests, `assert True` / `assert False` patterns, mock-only functions with no behavior verification
- **Test Binding Score**: Calculates a mutation resilience rating for your entire test suite (0–100%)
- **Mutation-verified repair**: Auto-generates replacement tests, then mutates source code to verify the new tests catch bugs before accepting them
- **Unconditional rollback**: All mutations are reverted in a `finally` block — your source code is never permanently mutated
- **Claude Code integration**: Works as a Claude Code skill (`/jmit`, `/jmitfix`) with zero extra API cost during repair
- **CI gate**: Blocks PRs that add non-binding tests
- **Platform support**: Works on all architectures (x86_64, aarch64) and operating systems where Python 3.11+ is available

## Technical Details

- **Language**: Python 3.11+
- **Test framework**: pytest
- **Installation**: `pip install jmit`
- **Claude Code alternative**: `/jmit` and `/jmitfix` skills (no install required)
- **Architecture**: Pure Python package (pip-installable), no binary required
- **Operating systems**: Windows, Linux, macOS
- **Mutation strategy**: Single-operator mutations (operator swap, boolean flip, return value change) applied one at a time, unconditionally rolled back

## Pricing

- **Free Trial**: 7 days, full access to `jmit run` and `jmit fix`, no credit card required. Starts automatically on first use.
- **Solo**: €29/month — 1 developer, full access
- **Team**: €99/month — up to 5 developers, full access
- **Enterprise**: Custom pricing — contact support@auroreci.com

All prices in EUR. Trial starts automatically; no manual activation required.

## Who JMIT Is For

- Python developers using pytest whose test suites have high coverage but still ship bugs
- Teams that want to enforce test quality in CI before merging
- Developers using Claude Code who want automated test repair inside their existing workflow
- Engineering leads who want objective, mutation-based test quality metrics (not just coverage %)

## Company

**Aurore Corporation International**
- Website: https://auroreci.com
- Support: support@auroreci.com
- Documentation: https://auroreci.com/docs

## Key Pages

- [Home](https://auroreci.com): Product overview and free trial signup
- [Documentation](https://auroreci.com/docs): Installation guide, CLI reference, repair protocol
- [Pricing](https://auroreci.com/pricing.html): Full plan comparison and feature matrix
- [Privacy Policy](https://auroreci.com/privacy.html): Data handling and privacy
- [Terms of Service](https://auroreci.com/terms.html): License terms and usage policy

## Frequently Asked Questions

**What is a zero-assertion test?**
A zero-assertion test is a test function that contains no `assert` statements. It will always pass regardless of what the code under test does, giving false confidence that the functionality works.

**What is the Test Binding Score?**
The Test Binding Score is JMIT's mutation resilience rating. It measures what percentage of JMIT-flagged weak tests were successfully repaired with mutation-verified binding tests. A score of 100% means every detected weak test now has a replacement that provably catches at least one class of bug.

**Does JMIT modify my source code permanently?**
No. All mutations applied during verification are unconditionally rolled back in a `finally` block. JMIT never permanently changes your source code — only test files are modified (to add or replace tests).

**What Python versions are supported?**
Python 3.11 and above.

**Does JMIT require a Claude API key?**
No. `jmit run` (audit) and `jmit fix` (repair) work without an API key. The Claude Code skill integration is optional and uses your existing Claude Code session at no extra API cost.

**How is JMIT different from coverage tools like pytest-cov?**
Coverage tools measure which lines of code are executed by tests. They do not measure whether tests actually verify the behavior of that code. A zero-assertion test achieves 100% line coverage while catching zero bugs. JMIT measures whether tests are *binding* — whether they would fail if the code were mutated.