Security audit
Security checks for vulnerabilities and agentic risk
This is a straightforward image-analysis skill that sends user-selected images and prompts to SiliconFlow, with privacy and cost considerations users should understand.
Install this only if you are comfortable sending selected images, image URLs, and prompts to SiliconFlow using your API key, with possible token charges. Avoid using it on confidential, regulated, biometric, or proprietary images unless you have permission and accept the provider's data-handling terms.
The skill uses environment variables and makes external network calls to a third-party vision API, but it does not declare an explicit tool scope such as allowed-tools or permissions. This weakens security boundaries and user visibility, making it easier for the skill to access sensitive inputs or transmit data externally without clear governance.
The skill encourages sending local images and image URLs to an external paid model/API but does not warn that image contents may contain sensitive personal, biometric, location, financial, or confidential business information. Because users may analyze private photos or internal screenshots, the omission increases the risk of unintentional data disclosure to a third party.
The skill is explicitly designed to transmit data to an external endpoint at api.siliconflow.cn, which means user-supplied prompts and image data leave the local environment. In the context of an image-analysis skill this behavior is expected, but it still represents a genuine data exposure risk if the tool is used on private, regulated, or proprietary content without adequate safeguards or transparency.
import urllib.request
import urllib.error
SILICONFLOW_API_URL = "https://api.siliconflow.cn/v1/chat/completions"
MODEL = "Qwen/Qwen2.5-VL-72B-Instruct"
The script sends either a local image (base64-embedded) or a remote image URL, along with the user prompt, to the third-party SiliconFlow API, but it provides no explicit disclosure or consent mechanism at the point of transmission. This is a real privacy/security issue because users may unknowingly transmit sensitive images, metadata-bearing URLs, or confidential prompts to an external service.
Skill's behavior or capabilities extend beyond its stated purpose. Scope creep allows an agent to perform actions unrelated to its documented functionality, increasing the attack surface.
permit persons to whom the Software is furnished to do so.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED,
INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR
PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE
FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
No suspicious patterns detected.