The Agent Skills Directory

[SAFE]: The skill demonstrates high security awareness by including a dedicated 'Security & Safety Notes' section. This section provides clear constraints to prevent the agent from inventing business facts (revenue, customer names) and warns against storing sensitive data like credentials or tax identifiers in the generated speaker notes. It also requires explicit authorization before external workspace edits.- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user input to generate structured data formats such as CSV, SQL DDL, and JSON. While it lacks specific instructions for sanitizing user-provided strings (e.g., escaping quotes for SQL), the risk is minimal as the skill focuses on content planning and the agent is instructed to adhere to a rigid schema and best practices. The ingestion points include audience details and business evidence, while the capabilities are limited to generating structured data templates for user review.

presentation-deck