Summary
This OpenAI walkthrough, presented by cyber deployment engineer Dan Shanahan, shows how to run a first security scan with Codex Security. It begins with confirming access to the Daybreak Blue model in Codex settings, then installing the Codex Security plugin and exploring its skills, including finding triage and deep scans.
The demo follows a full scan of a locally staged repository. Shanahan selects the model and reasoning level, enables a deep scan, and adds extra context to focus the analysis on the application’s authentication service. The scan builds a threat model, reviews code for candidate issues, and produces a validated list of vulnerabilities, each with a summary, attack path, and supporting evidence. He then generates a fix in a new Codex thread, reviews the proposed authorization changes, and publishes the patch as a GitHub pull request for the normal review and merge process.
The final section covers scaling up. The Codex Security CLI, installed through npm, runs the same scan steps from the command line, which makes it portable enough to embed in CI/CD pipelines and other automation across many repositories. The video is a practical starting point for security and engineering teams who want AI-assisted vulnerability discovery and remediation in their development workflow.
📺 Source: OpenAI · Published November 11, 2026
🏷️ Format: Tutorial Demo
