Security audit
Security checks for vulnerabilities and agentic risk
This skill is a Baidu video AI-notes wrapper whose network/API-key behavior is disclosed and aligned with its purpose, with only minor documentation clarity issues.
Install only if you are comfortable sending the video URL, the referenced video content, task IDs, and your Baidu API credential to Baidu's Qianfan service. Avoid using private, internal-only, confidential, or access-controlled video URLs unless that disclosure is acceptable under your policies. The docs have minor naming/help-text mistakes, but no hidden persistence or unrelated local access was found.
The skill tells users to provide a video URL but does not clearly warn that the referenced content will be fetched and processed by Baidu's external service. This can lead to unintended third-party data disclosure, privacy issues, or processing of sensitive/internal URLs that users may not realize are being transmitted outside the local environment.
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
params = {
"url": video_url,
}
response = requests.post(url, headers=headers, json=params)
response.raise_for_status()
result = response.json()
if "code" in result:
This code performs an outbound HTTP POST to a third-party Baidu endpoint using the supplied video URL and a bearer token from the environment. The script provides no confirmation prompt, disclosure message, or explanatory comment warning that user-supplied data will be transmitted off-system.
The skill is consistently described as an AI notes tool for videos, and the workflow states that the second API is AINotesTaskQuery. However, the heading at L45 says PPTOutlineGenerate API, which contradicts the documented script and parameter purpose that query an AI notes task by task ID. This is an intent/documentation contradiction rather than mere omission.
This code performs an HTTP request to an external Baidu endpoint using the provided task_id and bearer token, but the file offers no comment, docstring, or user-facing disclosure that task data will be transmitted off-host. Because the script is a query tool rather than an obviously network-centric wrapper with documented behavior, users may not be adequately warned about the external call.
The script implements a query operation for an existing AI notes task by accepting a task_id and calling the query endpoint at L08/L31, but the usage string says python ai_notes_task_create.py <url>, which describes a different script and a different kind of argument. This is an active documentation contradiction rather than a mere omission, and could mislead users about the skill's intended behavior.
No suspicious patterns detected.