wordpress-site-dna

Pass

Audited by Gen Agent Trust Hub on May 18, 2026

Risk Level: SAFEDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection as it processes untrusted content from the WordPress installation.
  • Ingestion points: The agent reads site content using wordpress_list_pages and wordpress_list_posts (SKILL.md).
  • Boundary markers: There are no instructions or delimiters provided to the agent to treat site content as untrusted data or to ignore embedded instructions.
  • Capability inventory: The agent has access to several audit tools (wordpress_get_site_info, wordpress_list_plugins, wordpress_get_database_stats) which could be misdirected to disclose technical site details (references/mcp-tools.md).
  • Sanitization: No sanitization or validation of the ingested content is specified.
  • [DATA_EXFILTRATION]: Evidence in the test suite indicates the skill performs data transmission that is not disclosed in the primary documentation.
  • Evidence: The file tests/benchmark.json includes pass criteria such as 'telemetry_sent: true' and 'telemetry POST fired' for standard audit tasks. This suggests the skill sends site audit data to a remote endpoint via POST requests during execution.
Audit Metadata
Risk Level
SAFE
Analyzed
May 18, 2026, 05:29 PM
Security Audit — agent-trust-hub — wordpress-site-dna