The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill facilitates the development of WordPress features that process external data, specifically post content, through AI-powered interfaces like wp_ai_client_prompt. This configuration creates a vulnerability surface for indirect prompt injection, where instructions embedded within a WordPress post could be interpreted as commands by the AI.
Ingestion points: The my_plugin_generate_summary_handler function in references/detailed-guide.md ingests post content.
Boundary markers: The guide includes explicit instructional warnings to "Treat post content as untrusted data, not instructions" and advises against automatic execution on critical hooks.
Capability inventory: The skill has the capability to interact with AI models and orchestrates several other technical skills for development and testing.
Sanitization: The provided code examples demonstrate basic input cleaning using strip_tags() and output sanitization via sanitize_textarea_field().

wordpress