The Agent Skills Directory

[SAFE]: The skill consists exclusively of informational Markdown files providing guidance on prompt engineering for various AI models. No active scripts, network operations, or unauthorized data access patterns were detected.
[INDIRECT_PROMPT_INJECTION]: While the skill provides templates for processing external data (e.g., data extraction in references/domain-data.md and code review in references/domain-code.md), it includes explicit safety guidance.
Ingestion points: Templates are designed to ingest external content such as code snippets, error messages, and raw text.
Boundary markers: The skill promotes the use of XML tags (e.g., <text>, <code>) and markdown delimiters to isolate external data from instructions.
Capability inventory: The documentation describes patterns for agents utilizing tools like search(), calculate(), and code() execution.
Sanitization: The skill includes a dedicated 'Safety Patterns' section in references/advanced-techniques.md covering jailbreak prevention, PII filtering, and output validation, which mitigates the risks associated with processing untrusted data.

ai-artist