The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user data (such as contract terms, party names, and scope) to generate a Word document. This creates a surface area for indirect prompt injection where instructions could be hidden within the provided data. However, the risk is minimal as the skill's capabilities are limited to document generation via platform tools and do not involve sensitive file access, network exfiltration, or shell command execution.
[SAFE]: The skill implements best practices for its use case by mandating legal disclaimers in both the output document and the user interaction, ensuring that the generated content is clearly identified as a reference draft rather than formal legal advice.

contract-drafter