alibabacloud-yike-storyboard

Pass

Audited by Gen Agent Trust Hub on Apr 17, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONDATA_EXFILTRATIONCREDENTIALS_UNSAFEPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses the official Aliyun CLI and a bundled bash script to automate storyboard job creation and monitor job status.
  • Evidence: SKILL.md and related-commands.md document the use of aliyun ice and aliyun ossutil commands.
  • [EXTERNAL_DOWNLOADS]: Fetches Aliyun CLI binaries from official vendor-managed distribution domains.
  • Evidence: cli-installation-guide.md points to aliyuncli.alicdn.com for tool downloads.
  • [REMOTE_CODE_EXECUTION]: Installs the required ice service plugin through the official Aliyun CLI plugin management system.
  • Evidence: SKILL.md includes instructions to run aliyun plugin install --names ice.
  • [DATA_EXFILTRATION]: Reads local script files and uploads them to Alibaba Cloud OSS as part of the core storyboard creation workflow. This behavior is documented and uses vendor-provided temporary credentials.
  • Evidence: scripts/upload_to_oss.sh performs the upload to vendor infrastructure using official STS tokens.
  • [CREDENTIALS_UNSAFE]: Manages API credentials via standard Aliyun CLI configuration and includes explicit warnings against exposing secret keys in responses or logs.
  • Evidence: SKILL.md contains a specific 'Security Rules' section for Access Key management.
  • [PROMPT_INJECTION]: Processes user-supplied text files, which presents a surface for indirect prompt injection in the downstream AI service.
  • Ingestion points: file_path (SKILL.md)
  • Boundary markers: Absent
  • Capability inventory: head (file read), aliyun ice (network API)
  • Sanitization: scripts/upload_to_oss.sh implements path traversal checks and file size limitations.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 17, 2026, 04:44 PM