baidu-cloud-bos

Pass

Audited by Gen Agent Trust Hub on Jun 16, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill's setup script downloads and installs the official @baiducloud/sdk package from the npm registry to provide Node.js functionality.
  • [COMMAND_EXECUTION]: The skill executes shell commands to utilize the bcecmd CLI tool and the local Node.js script for managing storage buckets and files.
  • [DATA_EXFILTRATION]: The skill facilitates the transfer of local data to external storage buckets on Baidu Cloud, which is its primary intended function.
  • [PROMPT_INJECTION]: The skill retrieves and processes object keys and metadata from external storage buckets, creating a surface for indirect prompt injection.
  • Ingestion points: Object lists and metadata are fetched from BOS via scripts/bos_node.mjs and bcecmd (e.g., during file listing or head operations).
  • Boundary markers: No specific delimiters or safety instructions are used to separate external storage content from agent instructions.
  • Capability inventory: The skill provides capabilities for file upload, download, deletion, and copying using local scripts and the bcecmd tool.
  • Sanitization: The skill does not implement validation or filtering for object keys or metadata retrieved from the remote storage service.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 16, 2026, 07:18 PM
Security Audit — agent-trust-hub — baidu-cloud-bos