Skip to content

Security: MHHamdan/Agentic-AI-Engineer

SECURITY.md

Security Policy

This repository is an open educational resource — it ships code samples, recipes, and labs, not a product with a deployed surface. Even so, a few things in here can be security-relevant: example code that handles secrets, recipes that demonstrate authentication, and security-focused content in security/. We take security issues in this material seriously and will respond promptly.


Supported versions

Because the repository's content evolves continuously, we support the current main branch only. Older tags are not maintained for security updates. If you depend on a specific snapshot of the content, pin it to a tag and update on your schedule.

Version Supported
main (latest)
Previous tagged releases

Reporting a vulnerability

Please report security issues privately, not as public issues or pull requests.

The preferred channel is GitHub's private security advisory feature:

➡️ Open a private security advisory

This is encrypted in transit and only visible to the repository maintainers.

If you can't use GitHub's advisory feature, contact @MHHamdan directly through a private GitHub message.

When reporting, please include:

  • A clear description of the issue and where it appears (file path, link, or recipe name).
  • A proof-of-concept or minimal reproduction if relevant.
  • The impact you observed or anticipate.
  • Whether the issue is in this repo's own content or in an upstream dependency.

We commit to:

  • Acknowledging your report within 7 days.
  • Providing a more detailed response within 14 days outlining next steps.
  • Crediting you in the fix (unless you prefer to remain anonymous).

What counts as a security issue in this repo

In scope Out of scope
Example code or recipes that demonstrate insecure patterns without warning Bugs that don't have security impact (open a regular issue)
Secrets accidentally committed in code, notebooks, or .env files Stylistic or content disagreements (open a regular issue)
CI workflows that could be abused to exfiltrate secrets or run untrusted code Vulnerabilities in dependencies (report to the upstream project, then optionally let us know)
Lab notebooks that, if run as-instructed, would create insecure systems Vulnerabilities in third-party tools we discuss (LangGraph, MCP servers, etc. — report upstream)
Compromise of the repository's supply chain Behavior of LLM providers (OpenAI, Anthropic, Google — report to those vendors)

Disclosing security-adjacent content in the repo

Some content in this repo intentionally discusses adversarial techniques — prompt injection, tool abuse, red-teaming, etc. We follow these principles when accepting such content:

  • Threat models and defenses with citations: yes.
  • Defensive patterns engineers can apply: yes.
  • Weaponizable exploits with no defensive purpose: no.
  • "100% prevention" claims for any class of attack: no.

See security/README.md and CONTRIBUTING.md for the full policy.


Coordinated disclosure for upstream issues

If you find an issue in an upstream dependency (e.g., a CVE in a library we recommend), please report it to the upstream project first, following their security policy. Once a fix is available or a CVE is assigned, let us know so we can update affected pages in this repo and add a migration note.


Acknowledgements

Security researchers who report valid issues will be acknowledged in CHANGELOG.md (with their consent), and optionally credited on the relevant fix's pull request.

Thank you for helping keep this resource safe.

There aren't any published security advisories