google-news-seo

Warn

Audited by Gen Agent Trust Hub on Mar 6, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes multiple shell commands to fetch and process web data, including curl, mktemp, sed, python3, and rm. These commands are used to download article HTML and manage temporary processing files.
  • [REMOTE_CODE_EXECUTION]: The skill employs a dynamic execution pattern where it generates a Python script (/tmp/extract_jsonld.py) at runtime using a heredoc and subsequently executes it. This script processes content retrieved from external, user-provided URLs.
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection (Category 8). It ingests untrusted data from external websites via web_fetch and curl to perform SEO and EEAT audits.
  • Ingestion points: External article HTML and JSON-LD metadata fetched from user-provided URLs in SKILL.md (Section 1 and 7).
  • Boundary markers: None detected; the agent is not instructed to ignore embedded instructions within the fetched HTML.
  • Capability inventory: The skill has the ability to execute shell commands (curl, python3, sed) and perform file system operations (mktemp, rm) in SKILL.md.
  • Sanitization: There is no evidence of sanitization or filtering of the fetched web content before it is processed by the agent or the dynamic Python script.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Mar 6, 2026, 12:08 PM