quality-e2e-testing

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill prompt embeds plaintext passwords/credentials in code and CI examples (e.g., POSTGRES_PASSWORD: testpass, test account passwords) which the LLM is expected to reproduce verbatim in its outputs, creating a secret-exfiltration risk.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.80). The GitHub Actions steps in the CI config fetch and execute external action repositories (actions/checkout@v4, actions/setup-node@v4, actions/upload-artifact@v4) at runtime in CI, which causes remote code to be downloaded and executed.