grade-tests

Provide the Skill with the list of changed test methods, the file paths containing them, or the PR diff. It resolves each test body, scores assertion strength, structure, and anti-patterns, then outputs a markdown table with letter grades and notes suitable for a PR comment.

It supports .NET (MSTest, xUnit, NUnit, TUnit), Python (pytest), TypeScript/JavaScript (Jest, Vitest), Java, Go, Ruby, Rust, Swift, Kotlin, PowerShell, and C++. Language-specific extension files define the assertion APIs and idiomatic patterns used during scoring.

Each test gets three sub-grades: assertion strength (45%), anti-pattern hygiene (30%), and structure (25%). The weighted average maps to a letter grade and score band, but the overall grade is capped at the worst sub-grade, so any F sub-grade forces an overall F.

Do not use it for full suite audits, writing new tests, fixing failures, or measuring code coverage. Use a test-quality-auditor or test-anti-patterns analysis for suite-wide reports, and a test-generation tool for authoring new tests.

The Skill rejects ambiguous requests with no explicit scope, such as "grade my tests" with no list. Provide an explicit list of fully-qualified test method names, file paths with an instruction to grade all tests in them, or a diff hunk identifying the changed tests.

No. Exception assertions such as Assert.ThrowsException, pytest.raises, or expect(fn).toThrow are treated as complete on their own. The Skill also counts mock verifications and bare assertion forms like pytest assert and Go if got != want checks as real assertions.