OpenAI 推出 AI 代码安全扫描 CLI/SDK，集检测、验证与修复于一体
Codex Security 是 OpenAI 推出的命令行工具和 TypeScript SDK，利用大语言模型（如 GPT-5.6-terra）扫描代码中的安全漏洞，并通过验证流程过滤误报，同时提供修复建议。它支持多种模型配置与努力等级，可集成至 CI/CD 流程，并按需使用 ChatGPT 登录或 API Key 认证。适合在开发阶段或部署前自动检测安全风险，降低人工审计成本。
OpenAI introduces @openai/codex-security, a CLI and TypeScript SDK that leverages AI to automatically find, validate, and fix security vulnerabilities in your code. It's designed for both local development and CI pipelines.
OpenAI推出了@openai/codex-security，这是一个命令行工具和TypeScript SDK，利用AI技术自动发现、验证和修复代码中的安全漏洞。它适用于本地开发环境以及持续集成流程。
Traditional scanners often suffer from high false-positive rates or missed vulnerabilities. @openai/codex-security uses OpenAI's code model to better understand code context and provide actionable fix recommendations.
传统安全扫描工具可能产生大量误报或漏报。@openai/codex-security利用OpenAI的代码模型，更准确地理解代码上下文，提供可操作的修复建议。
It supports both ChatGPT sign-in and API key authentication. You can specify the model (e.g., gpt-5.6-terra) and scan effort. Scan history is stored in a state directory, which can be overridden via environment variables.
支持ChatGPT登录和API密钥两种认证方式，可指定模型（如gpt-5.6-terra）和扫描力度（effort）。扫描历史保存在状态目录，支持环境变量覆盖。
Requires Node.js 22.13+ (22.x, 24.x, or 26.x) and Python 3.10+. Install via npm, then login and scan the current directory. Example commands:
npx @openai/codex-security login
npx @openai/codex-security scan .
Optionally specify model and effort:
npx @openai/codex-security scan . --model gpt-5.6-terra --effort high
安装需要Node.js 22.13+或24.x/26.x，Python 3.10+。使用npm安装，登录后即可扫描当前目录。命令示例：
npx @openai/codex-security login
npx @openai/codex-security scan .
可指定模型和力度：
npx @openai/codex-security scan . --model gpt-5.6-terra --effort high
@openai/codex-security ships as both a CLI and a TypeScript SDK. The CLI is ideal for ad-hoc scans and CI integration, while the SDK lets developers programmatically embed scans into their own toolchains. Both leverage OpenAI's code security model for analysis.
@openai/codex-security提供命令行工具和TypeScript SDK两种使用方式。CLI适合即时扫描和CI集成，SDK允许开发者以编程方式集成到自己的工具链中。两者都调用OpenAI的代码安全模型进行分析。
Ideal for local code security audits and CI/CD pipeline integration. Note: requires internet access to OpenAI's service, and a Trusted Access account is recommended for best results. Scan speed and cost may vary based on model and code size.
适合本地代码安全审计、CI/CD流水线集成。注意：需要网络连接到OpenAI服务，且建议使用Trusted Access账号以获得最佳效果。此外，扫描速度和成本可能因模型和代码量而异。