The Agent Skills Directory

Prompt Injection (SAFE): The skill contains instructional emphasis using keywords like 'CRITICAL' and 'MANDATORY' to ensure the agent maintains its '2026' persona. These are benign and do not attempt to bypass core safety filters or extract system prompts.\n- Data Exposure & Exfiltration (SAFE): No hardcoded credentials, sensitive file paths, or unauthorized network operations were detected. Code snippets for analytics (GA4) use standard placeholders like 'G-XXXXXXXXXX'.\n- Obfuscation (SAFE): No multi-layer obfuscation, Base64-encoded commands, homoglyphs, or zero-width characters were found in the analyzed files.\n- Remote Code Execution & Unverifiable Dependencies (SAFE): The bash and JavaScript blocks found in the documentation are provided as static templates for educational purposes. There is no logic for automated downloading or execution of untrusted scripts.\n- Indirect Prompt Injection (SAFE): While the skill provides guidelines for influencing AI search results (GEO), it does not contain executable tools that ingest untrusted data at runtime. The content is purely informational and represents a negligible security risk.

seo