zhipu-video-generation

Warn

Audited by Gen Agent Trust Hub on Sep 3, 2026

Risk Level: MEDIUMCREDENTIALS_UNSAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
  • [CREDENTIALS_UNSAFE]: The skill instructs the agent to load the ZHIPU_API_KEY from ~/.glm/.env. Accessing hidden files in the home directory is a sensitive operation that increases the risk of credential exposure if the agent's context is compromised.
  • [EXTERNAL_DOWNLOADS]: The Python SDK example uses from zai import ZhipuAI. The official SDK for Zhipu AI (BigModel) is published under the package name zhipuai. The use of zai is an unverifiable dependency and potentially a typosquatting attempt targeting users of Zhipu AI services.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied text to generate video content via API calls.
  • Ingestion points: The prompt parameter in the client.videos.generations.create call.
  • Boundary markers: None observed in the instructions or example code.
  • Capability inventory: Perform network POST requests to open.bigmodel.cn.
  • Sanitization: No input validation or sanitization logic is provided to handle malicious instructions embedded in the user prompt.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Sep 3, 2026, 07:15 PM
Security Audit — agent-trust-hub — zhipu-video-generation