OpenAI, 코드 보안 점검 도구 'Codex Security' 오픈소스 공개

Codex Security Codex Security is an open-source CLI and TypeScript SDK for finding, validating, and reviewing security issues in code you own or have permission to assess. Note This package follows semantic versioning. Its public API may change between minor versions before 1.0.0 . Requirements The SDK and CLI support macOS, Linux, and Windows and require Node.js 22 or later. Scanning and exporting findings also require Python 3.10 or later. If you use Python 3.10, install the tomli package. Python is not needed to install the package or run --help and --version . Sign in with your OpenAI account or provide an OpenAI API key before running a scan. Scan only repositories you own or have explicit permission to assess. Install and scan npm install @openai/codex-security npx codex-security login npx codex-security scan /path/to/repo Run npx codex-security --help to see all commands and npx codex-security scan --help for scan options. On a remote or headless machine, use npx codex-security login --device-auth . For CI and other unattended scans, set OPENAI_API_KEY or CODEX_API_KEY using your shell, CI secret, or secret manager. On Windows, set an API key in PowerShell with: $ env: OPENAI_API_KEY = " <your-api-key> " npx codex - security scan C:\code\repository To store an API key, pass it on stdin: printenv OPENAI_API_KEY | npx codex-security login --with-api-key Use npx codex-security login status to check the stored sign-in and npx codex-security logout to remove it. Codex Security reuses an existing file-based Codex sign-in. If Codex stores credentials in the system keyring, run npx codex-security login once before scanning. An environment API key takes precedence over a stored sign-in. Unset both OPENAI_API_KEY and CODEX_API_KEY to use your ChatGPT sign-in. The login status command reports the effective credential source without printing its value, including when no stored sign-in exists. Scan a subset of a repository or write machine-readable results: npx codex-security scan /path/to/repo --model gpt-5.6-terra npx codex-security scan /path/to/repo --path src --path tests npx codex-security scan /path/to/repo --knowledge-base /path/to/threat-models --knowledge-base /path/to/architecture.pdf npx codex-security scan /path/to/repo --diff origin/main --json npx codex-security scan /path/to/repo --output-dir /path/outside/repo/results npx codex-security scan /path/to/repo --output-dir /path/outside/repo/results --archive-existing npx codex-security scan /path/to/repo --dry-run npx codex-security scan /path/to/repo --fail-on-severity high npx codex-security install-hook npx codex-security bulk-scan npx codex-security bulk-scan repositories.csv --output-dir /path/outside/repositories/security-scans npx codex-security scans list /path/to/repo npx codex-security scans list --scan-root /path/outside/repo/results npx codex-security scans show SCAN_ID npx codex-security scans rerun SCAN_ID npx codex-security scans match PREVIOUS_SCAN_ID CURRENT_SCAN_ID npx codex-security scans match --all npx codex-security scans compare PREVIOUS_SCAN_ID CURRENT_SCAN_ID npx codex-security export /path/outside/repo/results --export-format sarif --output /path/outside/repo/results.sarif npx codex-security export /path/outside/repo/results --export-format csv --output /path/outside/repo/findings.csv npx codex-security export /path/outside/repo/results --export-format json --output /path/outside/repo/findings.json npx codex-security validate /path/outside/repo/findings.json " Possible SQL injection in src/query.ts:42 " npx codex-security patch /path/outside/repo/findings.json " Missing authorization check in src/routes.ts:18 " install-hook scans staged and unstaged changes before each commit. It respects core.hooksPath , does not replace an existing hook, and blocks high-severity findings or failed scans. Set --fail-on-severity to change the threshold. Use npx codex-security --version for the CLI version and npx codex-security info --json for package, plugin, and runtime versions, the default model and reasoning effort, and the next scan command. Add --dry-run to inspect the effective model and reasoning effort without initializing Codex or contacting the network. The output directory must be outside the scanned directory and any enclosing Git worktree. On macOS and Linux, an existing output directory must be private to the current user ( chmod 700 ). Scan artifacts can contain source excerpts, vulnerability details, and reproduction steps. Keep them out of repositories, public issue reports, and shared locations. When SARIF is produced, it is written to <scan-dir>/exports/results.sarif . Use npx codex-security scan --help for all target, output, and runtime options. Repeat --knowledge-base PATH for multiple files or directories. Directories are searched recursively for Markdown, text, PDF, and Word ( .docx ) files. Sign in with gh auth login , then run npx codex-security bulk-scan to discover GitHub repositories pushed in the last 90 days. Archived repositories and forks are excluded. Search the repository list, select the repositories to scan, and confirm before scanning. Private checkouts reuse your GitHub CLI sign-in without changing your global Git configuration. For automation or an existing repository list, pass a CSV containing id , repository , and full immutable revision columns and specify --output-dir . Use npx codex-security bulk-scan --help for all options. The CLI uses Incur for agent-friendly discovery and structured output. Use --llms for the command manifest, scan --schema --format json for a command schema, register an MCP server with mcp add , sync agent skills with skills add , and use completions bash|zsh|fish for shell completions. Scan results support --format toon|json|yaml|jsonl and --full-output . Use info --json for SDK and bundled-plugin metadata. MCP exposes only this read-only metadata command; scans, authentication, exports, validation, and patching remain CLI-only because the MCP transport cannot cancel active scans. If the output directory already contains results, add --archive-existing . The CLI moves them to <output-dir>.previous-<timestamp>-<id> and starts the scan in a new, empty directory at the original path. Add --dry-run to see the destination without moving files. Scans are report-only by default. Use --fail-on-severity in CI to exit 1 when a completed scan contains a finding at or above the selected severity. Incomplete coverage and CLI/runtime errors exit 2. Incomplete scans still write the available human or JSON result to stdout and a coverage warning to stderr, including in report-only mode. For CI, save machine-readable output outside the checked-out repository and apply a severity policy. Incomplete coverage and runtime errors still exit nonzero: SCAN_ROOT= " $( mktemp -d ) " npx codex-security scan . \ --diff origin/main \ --output-dir " $SCAN_ROOT /results " \ --json \ --fail-on-severity high > " $SCAN_ROOT /findings.json " JSON scans remain noninteractive, including when stderr is a terminal. Commands that run Codex interactively ( validate , patch , login , and logout ) reject --json . Write CSV exports to a file when JSON output is selected. Scans use gpt-5.6-sol with extra-high reasoning effort by default. Switch models with --model . Use --codex for other Codex settings: npx codex-security scan . --model gpt-5.6-terra --codex ' model_reasoning_effort="high" ' Scans report their requested paths and actual ranking, file-review, validation, and attack-path phases. Completion shows finding severity, coverage, elapsed time, available token and worker counts, the results directory, and the next useful command. Progress remains on stderr; JSON results remain on stdout. Use the TypeScript SDK Create a client, choose a private output directory outside the repository, and close the client after the scan: import { CodexSecurity } from "@openai/codex-security" ; const security = new CodexSecurity ( ) ; try { const result = await securit