wp-site-health-auditor

Pass

Audited by Gen Agent Trust Hub on Aug 17, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill generates instructions for significant system modifications, including editing critical files (wp-config.php, .htaccess), changing file permissions (chmod), and managing database content (wp search-replace). These capabilities are balanced by mandatory safety instructions requiring file backups and syntax validation (php -l) before application.
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and parse untrusted data from WordPress Site Health reports provided by the user as text or screenshots.
  • Ingestion points: Phase 1 (SKILL.md) describes parsing Status and Info reports from the WordPress admin dashboard.
  • Boundary markers: The skill does not define specific delimiters for the input data, but it enforces a strict "Phase 3 — Fix recipes by item" lookup logic to ensure the agent only performs actions explicitly defined in its catalog.
  • Capability inventory: Includes shell access for file manipulation, WP-CLI for database and plugin management, and system package management (apt install).
  • Sanitization: The skill relies on template-based matching against known WordPress health item titles to prevent the agent from executing arbitrary instructions embedded in a report.
  • [PRIVILEGE_ESCALATION]: The skill provides instructions for using sudo to install PHP extensions (references/catalog.md) and suggests specific file permission hardening (chmod 600), which are standard administrative tasks for the skill's purpose.
  • [PERSISTENCE_MECHANISMS]: The skill facilitates the creation of system cron jobs (references/catalog.md) to resolve WordPress background processing failures, which is a legitimate use case for persistence in a server management context.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 17, 2026, 07:42 AM
Security Audit — agent-trust-hub — wp-site-health-auditor