Skip to content

Security: Cyberzal1/Project-VEDA

Security

SECURITY.md

Security Policy

Scope of this project

Project VEDA is an autonomous, human-in-the-loop-governed red-team framework intended for authorised security testing and education only. It is designed to operate exclusively against targets you own or are explicitly permitted to test, inside the isolated PTOLEMAIOS sandbox.

The Singa Bank target is intentionally vulnerable

singa-bank-backend/ and singa-bank-frontend/ implement a deliberately vulnerable banking API (analogous to OWASP Juice Shop). Its flaws are seeded on purpose and flag-gated (off by default) for research and benchmarking. They are not security bugs to report — do not deploy this app on a public or production network.

Keys and secrets are not included

This repository intentionally ships no private keys or credentials:

  • The Ed25519 scope-gate keypair is generated locally by python setup_scope.py (written to keys/, which is git-ignored). Never commit *.key or the signed scope_manifest.json.
  • The target's JWT keys and any LLM API keys are supplied via environment variables (see singa-bank-backend/.env.example). Never commit real .env files.

Reporting a vulnerability in VEDA itself

If you find a security issue in the framework (not the intentionally vulnerable target), please open a GitHub issue marked "security", or contact the maintainer via the profile at https://github.com/Cyberzal1. Please do not include working exploits against third parties.

There aren't any published security advisories