The Agent Skills Directory

[SAFE]: The skill is designed for data governance and metadata management within the Amplitude platform. It does not perform arbitrary code execution or unauthorized network access.
[COMMAND_EXECUTION]: The skill utilizes specific tools for managing Amplitude taxonomy (e.g., set_event_description, hide_event). These tools are scoped to metadata operations and are governed by a strict requirement for explicit user confirmation before any changes are applied, as stated in the 'Core Philosophy' and 'Layer 2' rules.
[DATA_EXFILTRATION]: While the skill reads project context and taxonomy metadata through tools like get_event_metadata and get_context, these operations are conducted within the platform's authenticated environment. No patterns for exfiltrating data to external or unverified domains were detected.
[PROMPT_INJECTION]: The instructions emphasize following established standards, maintaining an 'evidence-first' stance, and explicitly warning against fabrication, which reinforces the agent's adherence to safety guidelines and intended behaviors.

taxonomy