ClawHub
Back to skill

Security audit

MiniMax Video Generation

Security checks for vulnerabilities and agentic risk

Overview

This skill is a straightforward MiniMax video-generation helper that sends chosen prompts or images to MiniMax and saves the generated video locally.

Install this only if you intend to use MiniMax cloud video generation. Do not submit confidential, regulated, or private images or prompts unless MiniMax processing is acceptable for your use case; keep the API key scoped and protected; and review the local output directory because generated videos remain on disk until removed.

SkillSpector

Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Taint TrackingDirect Taint Flow, Variable-Mediated Taint Flow, Credential Exfiltration Chain
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (15)

Tainted flow: 'download_url' from requests.get (line 114, network input) → requests.get (network output)

Medium
Category
Data Flow
Content
download_url = result["file"]["download_url"]

    print(f"正在下载视频 (约 3-5MB)...\n")
    video_resp = requests.get(download_url, timeout=180)
    with open(save_path, 'wb') as f:
        f.write(video_resp.content)
Confidence
89% confidence
Finding
The code blindly trusts a download_url returned by a remote API and performs a second requests.get to that URL without validating the scheme, host, or redirect behavior. If the upstream service is compromised, misconfigured, or attacker-controlled, this can turn the client into a server-side request gadget to fetch unexpected resources or download maliciously large content.

Lp3

Medium
Category
MCP Least Privilege
Confidence
95% confidence
Finding
The skill documents use of environment variables and outbound network access to a third-party API, but no corresponding permissions are declared. This creates a transparency and governance gap: users and hosting platforms may not realize the skill can read secrets and transmit prompts or images externally.

Missing User Warnings

Medium
Confidence
94% confidence
Finding
The workflow states that videos are automatically downloaded and saved locally, and examples upload prompts/images to an external service, but the documentation does not warn about third-party data transmission or privacy implications. Users may unknowingly send sensitive images or text to an external vendor and persist generated media on disk without informed consent.

Missing User Warnings

Medium
Confidence
90% confidence
Finding
The script sends user prompts and, when provided, raw local image contents to a third-party API without any explicit consent prompt, privacy notice, or data minimization controls. In an agent-skill context, users may assume local files stay local, so the lack of disclosure increases the risk of unintended sensitive-data transmission.

External Transmission

Medium
Category
Data Exfiltration
Content
API_KEY = os.getenv("MINIMAX_API_KEY")

response = requests.post(
    "https://api.minimaxi.com/v1/video_generation",
    headers={
        "Authorization": f"Bearer {API_KEY}",
Confidence
92% confidence
Finding
The documented POST request to the MiniMax API is a real external transmission of user content and authenticated metadata. Although central to the skill's function, it remains a true security/privacy concern because prompts may contain personal or confidential data and are sent outside the local environment.

External Transmission

Medium
Category
Data Exfiltration
Content
with open("start_image.jpg", "rb") as f:
    img_base64 = base64.b64encode(f.read()).decode()

response = requests.post(
    "https://api.minimaxi.com/v1/video_generation",
    headers={"Authorization": f"Bearer {API_KEY}", "Content-Type": "application/json"},
    json={
Confidence
96% confidence
Finding
The image-upload POST request transmits local file contents to an external provider. In a media-generation skill this is expected, but it materially increases risk because local images may contain highly sensitive personal or business information.

External Transmission

Medium
Category
Data Exfiltration
Content
### 首尾帧生成视频

```python
response = requests.post(
    "https://api.minimaxi.com/v1/video_generation",
    headers={"Authorization": f"Bearer {API_KEY}", "Content-Type": "application/json"},
    json={
Confidence
88% confidence
Finding
The POST request for start/end frame generation sends data to a third-party service, which is an actual external transmission. The context makes it functionally necessary, but users still need to understand that their image references and prompts leave the system.

External Transmission

Medium
Category
Data Exfiltration
Content
API_KEY = os.getenv("MINIMAX_API_KEY")

response = requests.post(
    "https://api.minimaxi.com/v1/video_generation",
    headers={
        "Authorization": f"Bearer {API_KEY}",
Confidence
92% confidence
Finding
The documented POST request to the MiniMax API is a real external transmission of user content and authenticated metadata. Although central to the skill's function, it remains a true security/privacy concern because prompts may contain personal or confidential data and are sent outside the local environment.

External Transmission

Medium
Category
Data Exfiltration
Content
with open("start_image.jpg", "rb") as f:
    img_base64 = base64.b64encode(f.read()).decode()

response = requests.post(
    "https://api.minimaxi.com/v1/video_generation",
    headers={"Authorization": f"Bearer {API_KEY}", "Content-Type": "application/json"},
    json={
Confidence
96% confidence
Finding
The image-upload POST request transmits local file contents to an external provider. In a media-generation skill this is expected, but it materially increases risk because local images may contain highly sensitive personal or business information.

External Transmission

Medium
Category
Data Exfiltration
Content
### 首尾帧生成视频

```python
response = requests.post(
    "https://api.minimaxi.com/v1/video_generation",
    headers={"Authorization": f"Bearer {API_KEY}", "Content-Type": "application/json"},
    json={
Confidence
88% confidence
Finding
The POST request for start/end frame generation sends data to a third-party service, which is an actual external transmission. The context makes it functionally necessary, but users still need to understand that their image references and prompts leave the system.

External Transmission

Medium
Category
Data Exfiltration
Content
API_KEY = os.getenv("MINIMAX_API_KEY")

response = requests.post(
    "https://api.minimaxi.com/v1/video_generation",
    headers={
        "Authorization": f"Bearer {API_KEY}",
        "Content-Type": "application/json"
Confidence
90% confidence
Finding
The hardcoded MiniMax API endpoint confirms third-party network egress. This is not inherently malicious, but it is a genuine privacy and supply-chain boundary crossing that should be treated as security-relevant in a skill that handles prompts and media.

External Transmission

Medium
Category
Data Exfiltration
Content
# 查询状态
status_resp = requests.get(
    f"https://api.minimaxi.com/v1/query/video_generation?task_id={task_id}",
    headers={"Authorization": f"Bearer {API_KEY}"}
)
status = status_resp.json()["status"]
Confidence
86% confidence
Finding
The status-query request sends a task identifier to the external service. This is lower impact than media upload, but it still reveals usage metadata and confirms ongoing communication with a third party.

External Transmission

Medium
Category
Data Exfiltration
Content
# 获取下载链接（有效期1小时，需立即下载）
    file_resp = requests.get(
        f"https://api.minimaxi.com/v1/files/retrieve?file_id={file_id}",
        headers={"Authorization": f"Bearer {API_KEY}"}
    )
    download_url = file_resp.json()["file"]["download_url"]
Confidence
91% confidence
Finding
The file retrieval call obtains a download URL from the external vendor, which extends the trust boundary and can lead to downloading externally hosted content into the local environment. This is expected in the workflow, but it adds risk because unverified remote content is then persisted locally.

External Transmission

Medium
Category
Data Exfiltration
Content
img_base64 = base64.b64encode(f.read()).decode()

response = requests.post(
    "https://api.minimaxi.com/v1/video_generation",
    headers={"Authorization": f"Bearer {API_KEY}", "Content-Type": "application/json"},
    json={
        "model": "MiniMax-Hailuo-2.3",
Confidence
96% confidence
Finding
This endpoint use accompanies the upload of a base64-encoded local image to the vendor API. The context makes the transmission intentional, but the risk is significant because private local content is leaving the user environment for third-party processing.

External Transmission

Medium
Category
Data Exfiltration
Content
```python
response = requests.post(
    "https://api.minimaxi.com/v1/video_generation",
    headers={"Authorization": f"Bearer {API_KEY}", "Content-Type": "application/json"},
    json={
        "model": "MiniMax-Hailuo-02",
Confidence
88% confidence
Finding
This endpoint invocation shows third-party submission for the start/end frame workflow. While aligned with the skill purpose, it still represents off-platform processing of user data and should not be treated as harmless by default.

Static analysis

No suspicious patterns detected.