semantic-core-builder

Pass

Audited by Gen Agent Trust Hub on Jul 29, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill includes shell commands in the Validation section of SKILL.md for skill maintenance and linting. These commands execute Python scripts (quick_validate.py, lint_production_skill.py) located within the $CODEX_HOME environment variable path.
  • [PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface as it is designed to process external, untrusted keyword and search-demand data.
  • Ingestion points: The agent is instructed to load reports/keyword-harvest.json and other user-provided keyword data.
  • Boundary markers: There are no explicit instructions to use delimiters or to disregard instructions that might be embedded within the keyword datasets.
  • Capability inventory: The skill allows the agent to write files to the local file system (reports/semantic-core.json, docs/site-structure.md) and references shell execution capabilities for validation.
  • Sanitization: The instructions do not specify any sanitization, filtering, or validation of the input keyword data before it is processed into semantic clusters.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 29, 2026, 07:30 AM
Security Audit — agent-trust-hub — semantic-core-builder