# Supercov

> Supercov measures code coverage for coding agents and software factories, giving AI agents actionable coverage gaps to write focused tests across JavaScript, TypeScript, Python, Ruby, and Rust.

Supercov is an open-source CLI tool built by Supercorp that gives coding agents—Claude Code, Codex, Cursor, Gemini CLI, GitHub Copilot, and others—actionable code coverage evidence to drive test writing. It runs the test suite you already have, records local coverage data, and surfaces uncovered paths as small, targeted queries that an agent can act on immediately. No account, config file, import, custom reporter, or hosted service is required.

## What It Is

Supercov sits between your existing test runner and your coding agent. You prefix your normal test command with `npx supercov --` (or the equivalent for Python, Ruby, or Rust), and Supercov instruments the run, stores immutable local evidence under `.supercov/`, and exposes CLI queries that return file-by-file coverage, uncovered lines, branches, MC/DC decisions, and assertion gaps. The agent reads those queries, writes one focused test, reruns the suite, and uses `supercov diff` to prove what improved. The loop continues while useful gaps remain.

## Coverage Depth

Supercov goes beyond line and branch coverage to include metrics that most tools omit:

- **Lines, statements, functions, and branches** — standard structural coverage
- **MC/DC (Modified Condition/Decision Coverage)** — independence witnesses for each condition in a decision
- **Optional-chain, default-value, and logical-assignment paths** — JavaScript/TypeScript-specific paths
- **`try`/`catch` and zero-iteration control-flow paths**
- **Per-test provenance** — exact attribution of which test covered which path, where the runner supports it
- **Assertion coverage** — tracks what the assertions in your tests actually check, separate from MC/DC

The denominator is computed from source structure before the run, so adding or removing tests cannot silently change what 100% means.

## Supported Languages and Test Runners

Supercov ships a single native binary through each language's own package manager at the same version:

- **JavaScript / TypeScript** — Jest, Vitest, Mocha, AVA, Playwright, `node:test`
- **Python** — pytest, unittest (CPython 3.12+)
- **Ruby** — RSpec, Minitest, test-unit, Cucumber (Ruby 3.3+)
- **Rust** — cargo test, cargo-nextest, rustdoc doctests
- **Go** — listed as coming later

Exact per-test attribution is available for Playwright, Vitest, Jest, `node:test`, RSpec, Minitest, test-unit, Cucumber, and Rust's libtest and cargo-nextest runners. Supercov works with Vite, Next.js, Turbopack, Webpack, esbuild, SWC, and projects with no build step.

## Agent Workflow

The intended loop is:

1. Run the real suite via `npx supercov -- <your test command>`
2. Query gaps with `npx supercov runs latest gaps --limit 10`
3. The agent writes one focused test targeting an uncovered path
4. Rerun the suite and prove the gain with `npx supercov diff <previous-run-id> latest`
5. Repeat while useful gaps remain

The homepage provides two ready-made prompts—one for writing a single useful test and one for a "software factory" loop that runs until coverage reaches 100%. These prompts can be pasted directly into any supported coding agent.

## Local, Private, and Zero-Edit

All data Supercov writes lives under a single `.supercov/` directory in the project. The CLI does not contact a Supercov service during a coverage run. It does not rewrite source files, tests, imports, reporter lists, runner configuration, dependency trees, or normal build output. Cleanup commands (`supercov clean`) let users preview and remove stored runs and the build cache.

## Update: v0.0.45

The latest release is v0.0.45, published on September 12, 2026. The repository was created in August 2026 and has been releasing frequently since launch. The project is MIT licensed, written primarily in Rust, and backed by Supercorp's open-source infrastructure. The homepage reports over 200k weekly npm downloads (vendor-published figure) and 3.4k GitHub stars across the Supercorp organization.

## Features
- MC/DC (Modified Condition/Decision Coverage) measurement
- Line, branch, statement, and function coverage
- Assertion coverage mapping
- Per-test provenance and exact attribution
- Uncovered path queries for coding agents
- Immutable local run evidence
- Coverage diff between runs
- Supports JavaScript, TypeScript, Python, Ruby, and Rust
- Works with Jest, Vitest, Playwright, pytest, RSpec, cargo test, and more
- No account, config file, or hosted service required
- Paginated CLI queries compatible with coding agent context windows
- Software factory loop support
- Low instrumentation overhead
- MIT licensed and fully open source

## Integrations
Claude Code, Codex, Cursor, Gemini CLI, GitHub Copilot, Jest, Vitest, Mocha, AVA, Playwright, node:test, pytest, unittest, RSpec, Minitest, test-unit, Cucumber, cargo test, cargo-nextest, rustdoc, Vite, Next.js, Turbopack, Webpack, esbuild, SWC, npm, PyPI, RubyGems, crates.io

## Platforms
WINDOWS, MACOS, LINUX, API, CLI

## Pricing
Open Source

## Version
v0.0.45

## Links
- Website: https://supercov.com
- Documentation: https://supercov.com/docs
- Repository: https://github.com/supercorp-ai/supercov
- EveryDev.ai: https://www.everydev.ai/tools/supercov