Code-X

How Code-X works

Submitted code is always treated as untrusted text. It is never executed - not in a shell, interpreter, browser, container, or sandbox.

1. Paste your code

Drop your source into the Monaco editor, select or auto-detect the language, and optionally describe what it's supposed to do.

2. Secrets are screened

Before anything is sent for review, the code is scanned for likely API keys, passwords, and credentials - submission is blocked until they're removed.

3. AI reviews the code

A configurable AI provider analyses correctness, security, code quality, maintainability, performance, error handling, and testing readiness against a weighted rubric.

4. Get a justified score

Receive a 0-10 score with a category breakdown, findings by severity, strengths, quick wins, and an improved version of your code.

5. Export and share

Download a PDF or HTML report, email it to yourself or a colleague, or create a revocable, expiring share link.

Weighted scoring rubric

Correctness & Reliability25%
Security20%
Code Quality & Readability15%
Maintainability15%
Performance & Efficiency10%
Error Handling & Resilience10%
Testing & Observability Readiness5%

A score of 10 is extremely rare and reserved for code with no meaningful issues. Critical security vulnerabilities materially reduce the overall score, regardless of whether the code otherwise runs.