OpenAI's Codex Security CLI and TypeScript SDK for finding, validating, and fixing security vulnerabilities. npm: https://www.npmjs.com/package/@openai/codex-security
TypeScript ai-securityapplication-securityclicode-scanningcodexcodex-securitycybersecuritydevsecopsnodejsnpmopenaisdksecuritytypescriptvulnerability-scanning
#!/usr/bin/env python3
"""Add the verified scanner payload to apko's single-platform SPDX document."""
import argparse
import base64
import hashlib
import json
from pathlib import Path, PurePosixPath
from urllib.parse import quote
def enrich(sbom, components):
"""Add installed npm packages, bundled tokenizer libraries, and native files to an SPDX document.
Args:
sbom (dict): apko's single-image SPDX document, updated in place.
components (dict): public-components.json inventory produced for the same image.
Returns:
dict: The input document with containment relationships and a new document namespace.
Raises:
ValueError: The image manifest in the inventory differs from the one in the SBOM.
"""
packages = sbom["packages"]
image, = (p for p in packages if p["SPDXID"] in sbom["documentDescribes"])
if image["name"] != components["image_manifest"]:
raise ValueError("apko SBOM and payload inventory describe different images")
apk, = (p for p in packages if p["name"] == "codex-security")
install = PurePosixPath("/usr/lib/codex-security")
def add(parent_id, package):
packages.append(package)
sbom["relationships"].append({"spdxElementId": parent_id, "relationshipType": "CONTAINS",
"relatedSpdxElement": package["SPDXID"]})
# Use the manifest name for npm aliases, but keep each installed path distinct.
npm_ids = {}
for i, item in enumerate(components["installed_npm"]):
path = PurePosixPath(item["path"])
npm_ids[path] = f"SPDXRef-npm-{i}"
name, version = item["name"], item["version"]
purl = f"pkg:npm/{quote(name)}@{quote(version, safe='')}"
checksum = base64.b64decode(item["integrity"].removeprefix("sha512-"), validate=True).hex()
add(apk["SPDXID"], {
"SPDXID": npm_ids[path], "name": name, "versionInfo": version,
"downloadLocation": item["source"], "licenseDeclared": item["license"],
"filesAnalyzed": False, "sourceInfo": f"Installed at {install / path}; SHA512 identifies the source archive.",
"checksums": [{"algorithm": "SHA512", "checksumValue": checksum}],
"externalRefs": [{"referenceCategory": "PACKAGE-MANAGER", "referenceType": "purl",
"referenceLocator": purl}],
})
for i, item in enumerate(components.get("bundled_tokenizer", [])):
add(npm_ids[PurePosixPath("node_modules/@openai/codex-security")], {
"SPDXID": f"SPDXRef-tokenizer-{i}", "name": item["name"], "versionInfo": item["version"],
"downloadLocation": "NOASSERTION", "licenseDeclared": item["licenses"][0]["license"]["id"],
"filesAnalyzed": False, "sourceInfo": "Bundled in the SDK embedding tokenizer.",
"externalRefs": [{"referenceCategory": "PACKAGE-MANAGER", "referenceType": "purl",
"referenceLocator": item["purl"]}],
})
for i, item in enumerate(components["bundled_native"]):
path = PurePosixPath(item["path"])
owner = next(npm_ids[p] for p in path.parents if p in npm_ids)
add(owner, {
"SPDXID": f"SPDXRef-native-{i}", "name": path.name, "primaryPackagePurpose": "FILE",
"packageFileName": str(install / path), "downloadLocation": "NOASSERTION",
"filesAnalyzed": False, "sourceInfo": item["source"],
"checksums": [{"algorithm": "SHA256", "checksumValue": item["sha256"]}],
})
# A changed SPDX document needs its own namespace, distinct from apko's original.
sbom["creationInfo"]["creators"].append("Tool: codex-security/release-sbom.py")
digest = hashlib.sha256(json.dumps(sbom, sort_keys=True).encode()).hexdigest()
sbom["documentNamespace"] = f"https://github.com/openai/codex-security/spdx/wolfi-{digest}"
return sbom
if __name__ == "__main__":
parser = argparse.ArgumentParser(description=__doc__)
parser.add_argument("sbom", type=Path)
parser.add_argument("components", type=Path)
parser.add_argument("destination", type=Path)
args = parser.parse_args()
result = enrich(json.loads(args.sbom.read_bytes()), json.loads(args.components.read_bytes()))
args.destination.write_text(json.dumps(result, indent=2) + "\n")