PANews

Pass

Audited by Gen Agent Trust Hub on Oct 24, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: Several workflows instruct the agent to interpolate user-provided keywords or extracted identifiers directly into shell command templates, which poses a risk of command injection if the agent does not properly escape shell metacharacters.
  • workflow-search.md: Instructions to execute node cli.mjs search-articles "<keyword>" using user-supplied keywords.
  • workflow-read-article.md: Instructions to extract an ID from a user-provided URL and execute node cli.mjs get-article <articleId>.
  • workflow-topic-research.md: Instructions to execute node cli.mjs search-articles "<user keyword>".
  • workflow-calendar.md, workflow-events.md, workflow-series.md, workflow-topics.md: All involve passing user search strings to shell commands.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes content from external sources (PANews articles and community comments) that could contain malicious instructions designed to influence the agent's behavior.
  • Ingestion points: Content is retrieved via get-article (full article text), get-topic (community comments), and search-articles (summaries).
  • Boundary markers: The workflows do not specify the use of delimiters (e.g., XML tags or triple quotes) or "ignore embedded instructions" warnings when processing the retrieved text.
  • Capability inventory: The skill has the capability to execute shell commands via the cli.mjs script.
  • Sanitization: There are no instructions provided to sanitize, filter, or escape the content retrieved from PANews before it is analyzed or summarized by the agent.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 24, 2026, 06:05 AM