alibabacloud-yike-storyboard
Pass
Audited by Gen Agent Trust Hub on Apr 28, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface because it reads and processes the first 1000 characters of user-supplied novel or script files to recommend visual styles and voice settings. 1. Ingestion points: User-provided text and docx files read in SKILL.md Task 0. 2. Boundary markers: No delimiters or instructions are used to distinguish user content from agent instructions. 3. Capability inventory: Shell command execution and file uploads via vendor CLI tools. 4. Sanitization: No content validation or instruction escaping is performed on the ingested text.
- [COMMAND_EXECUTION]: The skill performs shell operations using the aliyun CLI and a local helper script (scripts/upload_to_oss.sh) to configure environment settings, install plugins, and interact with cloud APIs.
- [EXTERNAL_DOWNLOADS]: The documentation instructs users to download the Aliyun CLI and service-specific plugins from the vendor's official distribution domains. These downloads are considered safe as they originate from the provider's official infrastructure.
Audit Metadata