article-processing

Warn

Audited by Gen Agent Trust Hub on Jun 20, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes multiple shell commands including npx, rg, aria2c, wget, and curl to process files and download content.\n- [COMMAND_EXECUTION]: User-provided or extracted URLs are directly interpolated into shell command arguments (e.g., npx uuid v5 "<source_identity_url>" URL). This presents a risk of command injection if the input strings are not properly escaped or sanitized by the agent.\n- [REMOTE_CODE_EXECUTION]: The skill uses npx to run the uuid package. npx can dynamically download and execute code from the npm registry, which introduces a dependency on external, third-party code at runtime.\n- [EXTERNAL_DOWNLOADS]: The skill automatically downloads images from external websites using aria2c, wget, or curl based on links found within articles.\n- [COMMAND_EXECUTION]: The skill uses rg (ripgrep) to perform searches in the local filesystem using variables that could be manipulated to perform directory traversal or unauthorized file reads.\n- [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface. \n
  • Ingestion points: Untrusted article content is extracted via the agent-browser tool (REFERENCE.md).\n
  • Boundary markers: None identified to delimit processed article data.\n
  • Capability inventory: Subprocess calls (npx, rg, curl, wget, aria2c) and file-write operations across all scripts.\n
  • Sanitization: No explicit sanitization or escaping of extracted content is specified before processing.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Jun 20, 2026, 02:51 PM
Security Audit — agent-trust-hub — article-processing