Security audit
Security checks for vulnerabilities and agentic risk
This skill provides local medical-device cybersecurity guidance and templates with no evidence of harmful execution or data transmission.
Use this as a reference and template starter. Confirm medical-device applicability, verify current regulatory requirements, and use a proper CVSS assessment before making remediation decisions. Review software-inventory contents and authorized recipients before sharing an SBOM. The verifier reads skill-package files and overwrites its local reports. Treat the bundled audit and ownership claims as unverified because of their inconsistencies. SkillSpector reported complete coverage; this assessment independently inspected all supplied files.
Without declared permissions the skill's intent is opaque and cannot be validated.
Step 6 instructs submission and supply-chain sharing of the SBOM, which contains component names, precise versions, suppliers, and dependency relationships. The file does not warn that this can disclose sensitive software inventory or instruct the agent to check recipient authorization before sharing.
subprocess module calls execute external commands. Without careful input validation, this enables command injection.
def check_tools():
items = []
try:
r = subprocess.run([PYTHON, TOOL, "reg", "--region", "us"], capture_output=True, text=True, timeout=60)
out = r.stdout or ""
items.append({"name": "reg 命令执行", "pass": r.returncode == 0 and "SBOM" in out and "RTA" in out})
r2 = subprocess.run([PYTHON, TOOL, "checklist", "--phase", "submission"], capture_output=True, text=True, timeout=60)
subprocess module calls execute external commands. Without careful input validation, this enables command injection.
r = subprocess.run([PYTHON, TOOL, "reg", "--region", "us"], capture_output=True, text=True, timeout=60)
out = r.stdout or ""
items.append({"name": "reg 命令执行", "pass": r.returncode == 0 and "SBOM" in out and "RTA" in out})
r2 = subprocess.run([PYTHON, TOOL, "checklist", "--phase", "submission"], capture_output=True, text=True, timeout=60)
items.append({"name": "checklist 命令执行", "pass": r2.returncode == 0 and "渗透测试" in (r2.stdout or "")})
r3 = subprocess.run([PYTHON, TOOL, "sbom"], capture_output=True, text=True, timeout=60)
out3 = r3.stdout or ""
subprocess module calls execute external commands. Without careful input validation, this enables command injection.
items.append({"name": "reg 命令执行", "pass": r.returncode == 0 and "SBOM" in out and "RTA" in out})
r2 = subprocess.run([PYTHON, TOOL, "checklist", "--phase", "submission"], capture_output=True, text=True, timeout=60)
items.append({"name": "checklist 命令执行", "pass": r2.returncode == 0 and "渗透测试" in (r2.stdout or "")})
r3 = subprocess.run([PYTHON, TOOL, "sbom"], capture_output=True, text=True, timeout=60)
out3 = r3.stdout or ""
items.append({"name": "sbom 命令执行", "pass": r3.returncode == 0 and "CycloneDX" in out3 and "components" in out3})
r4 = subprocess.run([PYTHON, TOOL, "vuln", "--desc", "远程可被利用执行任意代码"], capture_output=True, text=True, timeout=60)
subprocess module calls execute external commands. Without careful input validation, this enables command injection.
r3 = subprocess.run([PYTHON, TOOL, "sbom"], capture_output=True, text=True, timeout=60)
out3 = r3.stdout or ""
items.append({"name": "sbom 命令执行", "pass": r3.returncode == 0 and "CycloneDX" in out3 and "components" in out3})
r4 = subprocess.run([PYTHON, TOOL, "vuln", "--desc", "远程可被利用执行任意代码"], capture_output=True, text=True, timeout=60)
items.append({"name": "vuln 命令执行", "pass": r4.returncode == 0 and "严重" in (r4.stdout or "")})
r5 = subprocess.run([PYTHON, TOOL, "standard"], capture_output=True, text=True, timeout=60)
out5 = r5.stdout or ""
subprocess module calls execute external commands. Without careful input validation, this enables command injection.
items.append({"name": "sbom 命令执行", "pass": r3.returncode == 0 and "CycloneDX" in out3 and "components" in out3})
r4 = subprocess.run([PYTHON, TOOL, "vuln", "--desc", "远程可被利用执行任意代码"], capture_output=True, text=True, timeout=60)
items.append({"name": "vuln 命令执行", "pass": r4.returncode == 0 and "严重" in (r4.stdout or "")})
r5 = subprocess.run([PYTHON, TOOL, "standard"], capture_output=True, text=True, timeout=60)
out5 = r5.stdout or ""
items.append({"name": "standard 命令执行", "pass": r5.returncode == 0 and "81001" in out5})
except Exception as e:
subprocess module calls execute external commands. Without careful input validation, this enables command injection.
items = []
for desc, expects in cases:
try:
r = subprocess.run([PYTHON, TOOL, "vuln", "--desc", desc], capture_output=True, text=True, timeout=60)
ok = r.returncode == 0 and any(e in (r.stdout or "") for e in expects)
items.append({"name": f"vuln「{desc[:14]}…」→{expects[0]}", "pass": ok})
except Exception as e:
The skill explicitly requires activation whenever any listed phrase matches, including questions about SBOM depth and customer-reported vulnerability patch deadlines. These questions also arise in ordinary software development, so they can incorrectly activate medical-device regulatory guidance without any medical-device context.
The report's note explicitly promises no scanning, yet check_security() traverses BASE, reads supported files, and searches their contents for sensitive-data patterns (L124–L135). This is a local package scan, not evidence of network scanning or exfiltration, but the unqualified assurance contradicts the implemented behavior.