Thanks for considering contributing.
BeforeShip is intended to be a practical, open-source launch gate for AI-built apps. Contributions should make it easier for solo founders to identify and fix real production risks before deploy.
- Add a detector for a common vibe-coded app failure
- Add examples of flawed/fixed apps
- Improve report language
- Add framework-specific checks
- Improve documentation
- Add tests for detectors
- Add MCP client setup examples
- Prefer deterministic checks over LLM-only judgement.
- Explain findings in plain language. Solo founders should understand why an issue matters.
- Include evidence. Findings should point to files, routes, headers, or commands.
- Generate minimal fix plans. Avoid broad “rewrite the app” advice.
- Do not encourage offensive testing against third-party apps. BeforeShip is for apps you own or are authorized to audit.
Each detector should return findings shaped like:
{
"id": "category.issue_id",
"severity": "critical",
"category": "payments",
"title": "Human-readable title",
"file": "path/to/file.ts",
"evidence": "What was found",
"why_it_matters": "Plain-English risk",
"fix_prompt": "Agent-ready fix prompt"
}Implementation is not started yet. Once packages are added, this file will include local setup, test, lint, and release instructions.
If you find a vulnerability in BeforeShip itself, please report it privately first. See SECURITY.md.