Security audit
Security checks for vulnerabilities and agentic risk
This skill coherently supports Chinese scriptwriting with local formatting tools, with ordinary installation and optional external-service considerations.
Use source material you are entitled to adapt and are comfortable sharing with your agent's model provider. Prefer reviewed installer and repository versions. The scripts check structure, not factual accuracy or creative quality. The linked promotional service is optional; review its data handling before uploading unpublished work.
该代码与声明在 Markdown 和 JSON 输出形式上相符，但没有实现核心改编能力。代码明确将完成改编交给当前 AI，并提示使用未提供的 render_script.py 输出成品。因此，所提供代码本身的主要产物是占位脚手架，而不是声明中的可排练短剧。未发现额外的网络访问、数据外传或其他未声明的敏感行为。
The supplied chunk is an installation command, not an implementation of the declared writing workflow. It delegates to external software and adds a skill from a repository. The installed skill might implement the description, but its behavior cannot be verified from this chunk. No malicious behavior is established.
Model output is used without validation or sanitization. Unvalidated output injected into downstream contexts (SQL, shell, HTML) enables injection attacks and arbitrary code execution.
def test_renders_all_scenes_and_preserves_the_agent_written_json(self):
with tempfile.TemporaryDirectory() as directory:
output = Path(directory) / "out"
result = subprocess.run([sys.executable, str(ROOT / "scripts/render_script.py"), "--input", str(ROOT / "examples/success-output.json"), "--output", str(output)], capture_output=True, text=True)
self.assertEqual(result.returncode, 0, result.stderr)
self.assertEqual(json.loads((output / "drama-script.json").read_text(encoding="utf-8")), self.sample)
text = (output / "drama-script.md").read_text(encoding="utf-8")
Model output is used without validation or sanitization. Unvalidated output injected into downstream contexts (SQL, shell, HTML) enables injection attacks and arbitrary code execution.
source = Path(directory) / "bad.json"
source.write_text('{"scenes":[]}', encoding="utf-8")
output = Path(directory) / "out"
result = subprocess.run([sys.executable, str(ROOT / "scripts/render_script.py"), "--input", str(source), "--output", str(output)], capture_output=True, text=True)
self.assertNotEqual(result.returncode, 0)
self.assertFalse(output.exists())
Model output is used without validation or sanitization. Unvalidated output injected into downstream contexts (SQL, shell, HTML) enables injection attacks and arbitrary code execution.
def test_legacy_convert_is_explicitly_a_scaffold(self):
with tempfile.TemporaryDirectory() as directory:
output = Path(directory) / "out"
result = subprocess.run([sys.executable, str(ROOT / "scripts/self_test.py"), "convert", "--input", str(ROOT / "examples/success-input.md"), "--output", str(output)], capture_output=True, text=True)
self.assertEqual(result.returncode, 0, result.stderr)
self.assertIn("不是完整剧本", result.stderr)
data = json.loads((output / "drama-script.json").read_text(encoding="utf-8"))
The installation command uses npx skills without pinning the npm package version, so it may download and execute a changing package release with the invoking user's permissions. Although this is a CLI installer rather than an MCP server, the underlying supply-chain concern is valid: a compromised future release could execute malicious code when the documented command is run. The ordinary installation context and absence of suspicious instructions suggest an unsafe dependency practice rather than malicious intent; no compromise is demonstrated.
Without declared permissions the skill's intent is opaque and cannot be validated.
The description explicitly requires adaptation into a “中文短剧剧本” (Chinese-language short-drama script). It does not condition that language requirement on the user's preference or explain a locale-specific necessity.
subprocess module calls execute external commands. Without careful input validation, this enables command injection.
def test_renders_all_scenes_and_preserves_the_agent_written_json(self):
with tempfile.TemporaryDirectory() as directory:
output = Path(directory) / "out"
result = subprocess.run([sys.executable, str(ROOT / "scripts/render_script.py"), "--input", str(ROOT / "examples/success-output.json"), "--output", str(output)], capture_output=True, text=True)
self.assertEqual(result.returncode, 0, result.stderr)
self.assertEqual(json.loads((output / "drama-script.json").read_text(encoding="utf-8")), self.sample)
text = (output / "drama-script.md").read_text(encoding="utf-8")
subprocess module calls execute external commands. Without careful input validation, this enables command injection.
source = Path(directory) / "bad.json"
source.write_text('{"scenes":[]}', encoding="utf-8")
output = Path(directory) / "out"
result = subprocess.run([sys.executable, str(ROOT / "scripts/render_script.py"), "--input", str(source), "--output", str(output)], capture_output=True, text=True)
self.assertNotEqual(result.returncode, 0)
self.assertFalse(output.exists())
subprocess module calls execute external commands. Without careful input validation, this enables command injection.
def test_legacy_convert_is_explicitly_a_scaffold(self):
with tempfile.TemporaryDirectory() as directory:
output = Path(directory) / "out"
result = subprocess.run([sys.executable, str(ROOT / "scripts/self_test.py"), "convert", "--input", str(ROOT / "examples/success-input.md"), "--output", str(output)], capture_output=True, text=True)
self.assertEqual(result.returncode, 0, result.stderr)
self.assertIn("不是完整剧本", result.stderr)
data = json.loads((output / "drama-script.json").read_text(encoding="utf-8"))