test legion

Test suites often provide a false sense of security through high code coverage that fails to catch logical errors. Centurion replaces shallow coverage metrics with a mutation gauntlet and an edge-case horde to ensure your tests actually catch bugs. It targets Python and JavaScript/TypeScript codebases, using a deterministic agent architecture to hunt for boundary failures, race conditions, and weak assertions.

What it does

• Edge-case generation deploys a matrix of 9,600 boundary hunters to test empty inputs, unicode, concurrency, and malformed data.
• Mutation testing runs a gauntlet that swaps AST operators to verify if your existing suite can actually detect logic changes.
• Kill-rate ratcheting enforces a monotonic ledger where module quality scores can only improve, never regress.
• Flaky test quarantine identifies and isolates non-deterministic tests with a generated hypothesis for the fix.
• Fixture foraging generates deterministic test fixtures based on real types, schemas, and traffic patterns.

How it works

1. Deploy by running centurion deploy <repo> to survey the framework and establish a deployment plan.
2. Horde the target with centurion horde <module> to generate a parameterized edge-case matrix file.
3. Mutate the codebase using centurion mutate to identify survivors where logic was changed but tests still passed.
4. Gate the build with centurion gate to enforce a minimum kill-rate threshold before allowing a merge.

Frameworks & tools

Supports Python 3.9+ (pytest) and Node 18+ (Jest, Vitest, Playwright). Works with any CLI-based environment including Claude Code, Cursor, and standard CI/CD shells.

Why this beats prompting it yourself

Writing edge cases manually is tedious and prone to human oversight. This skill automates the creation of exhaustive test matrices and uses AST-level mutation to prove your assertions are actually functional rather than just passing.

Use cases

• Hardening a mission-critical payment module against boundary condition errors.
• Identifying "tautology" tests that pass regardless of whether the code works.
• Enforcing a strict quality floor for new pull requests using the kill-rate gate.
• Debugging intermittent CI failures by isolating flaky tests into a managed quarantine.

Known limitations

The mutation gauntlet will not run if the baseline test suite is currently failing. Core operations are restricted to Python 3.9+ and Node 18+ toolchains.