AI Engineering Lab is a training repository: it contains educational notebooks and guides, not production services. Still, learners execute this code with real credentials, so we take a few things seriously.

If you find something that could put a learner's cloud account, API key, or data at risk, a notebook pattern that leaks secrets, an insecure default in a guide, please report it privately to info@zorost.com (or via GitHub's private vulnerability reporting). We will acknowledge within 5 business days.

1. No secrets in the repo. Never commit API keys, tokens, or workspace URLs. Use the .env.example pattern; each notebook documents which variable it needs.
2. Least privilege. Cloud setup guides (Weeks 18 to 24) always start from minimal IAM roles / scoped permissions, never admin.
3. Blast radius. Agent guides (Weeks 12 to 17) mark commands that touch production data, spend money, or are irreversible with a ⚠️ banner.
4. Untrusted content. Prompt-injection and jailbreak material is educational and always paired with defenses (knowledge-base 04).
5. Cost guardrails. Every paid-platform guide tells you to set a budget alert before running anything.

If you add a notebook that calls a paid API, it must (a) read the key from an environment variable, (b) print an estimated cost before running heavy loops, and (c) include a kill switch note.

© 2026 Zorost Intelligence LLC · https://zorost.com