The Agent Skills Directory

[SAFE]: The skill is entirely educational, providing templates and best practices for developing software that integrates with AI APIs. The code examples are transparent and follow industry standards for security and resource management.
[INDIRECT_PROMPT_INJECTION]: The skill documents a data ingestion surface where user input is processed by an LLM. It mitigates this risk by recommending an architecture that includes 'Input Validation + Sanitization' as a mandatory first step and providing implementation examples that enforce input length constraints.
Ingestion points: User input string in the generateContent function in SKILL.md.
Boundary markers: The 'AI Product Architecture' diagram explicitly includes validation and sanitization layers before prompt construction.
Capability inventory: The skill uses standard API calls to the Anthropic SDK to generate text responses.
Sanitization: A length check is implemented in the provided Javascript example (userInput.length > 5000).
[EXTERNAL_DOWNLOADS]: The skill references the official @anthropic-ai/sdk library. This is a well-known service used for legitimate integration with AI models, and its inclusion is consistent with the skill's primary purpose.

ai-wrapper-product