wordpress

Pass

Audited by Gen Agent Trust Hub on Jun 16, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill describes an implementation of an AI-powered content summarizer that is vulnerable to Indirect Prompt Injection.
  • Ingestion points: The save_post hook example in SKILL.md (lines 316-353) reads raw post_content from the database to send to an AI client.
  • Boundary markers: Absent. The instruction wp_ai_client_prompt('Create a brief 2-sentence summary of this content: ' . substr($content, 0, 1000)) interpolates untrusted data directly without delimiters (like ### or XML tags) or instructions to ignore instructions embedded in the content.
  • Capability inventory: The skill demonstrates the ability to write to the database (wp_update_post) based on AI output, which could allow an attacker to manipulate site data via injected instructions.
  • Sanitization: The example uses strip_tags() and sanitize_textarea_field(); while these protect against HTML/XSS, they do not mitigate adversarial instructions within the text content.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 16, 2026, 01:00 AM
Security Audit — agent-trust-hub — wordpress