blog-write

Pass

Audited by Gen Agent Trust Hub on Mar 30, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill is subject to indirect prompt injection (Category 8) risks because it automatically gathers and processes data from external web searches and websites.
  • Ingestion points: Web search results and fetched page content from various sources during the research phase.
  • Boundary markers: Absent. The instructions do not define delimiters to separate untrusted external data from the agent's core logic.
  • Capability inventory: Bash (for verification), Write (for article drafting), Edit (for refining content), and WebFetch (for retrieving external data).
  • Sanitization: Absent. There is no specific requirement for the agent to filter or escape content before including it in the generated blog post.
  • [COMMAND_EXECUTION]: In Phase 2.3, the skill instructs the agent to execute bash commands (curl -sI) using URLs found during web searches. If a search result provides a malicious URL containing shell metacharacters, it could potentially lead to command injection if the underlying tool interface does not perform adequate sanitization.
  • [EXTERNAL_DOWNLOADS]: The skill programmatically references and validates images from well-known services such as Pixabay, Unsplash, and Pexels. These downloads are part of the intended functionality and target reputable platforms.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 30, 2026, 02:44 AM