keyword-research

Pass

Audited by Gen Agent Trust Hub on Aug 20, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands using python3 for local connector scripts (suggest.py, firecrawl.py, pageviews.py). These instructions include placeholders like <seed> and <candidate keyword> intended for user-supplied data. This creates a potential command injection surface if the agent interpolates these values without proper shell escaping.
  • [PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection as it ingests and processes data from various external sources.
  • Ingestion points: Seed keywords provided by the user, search suggestions from Google, SERP results from Firecrawl, and traffic metrics from Wikipedia.
  • Boundary markers: None specified; the instructions do not require the use of delimiters or 'ignore embedded instructions' prompts when handling the external content.
  • Capability inventory: The skill possesses the capability to execute shell commands through the Python interpreter and write research reports to the local memory/ directory.
  • Sanitization: The instructions do not describe any mechanisms to sanitize, validate, or filter content retrieved from the external web endpoints.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 20, 2026, 03:37 PM
Security Audit — agent-trust-hub — keyword-research