The Agent Skills Directory

[SAFE]: The skill consists entirely of instructional Markdown content and contains no executable scripts, binaries, or active code components. This lack of executable code significantly limits the risk of direct execution or privilege escalation attacks.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze business data (such as dealer metrics) to generate visual insights. While this creates an ingestion point for untrusted data, the skill mitigates this risk by establishing mandatory logical gates, intent validation, and North Star metric identification, which act as a structured framework for sanitizing the agent's reasoning process.
[DATA_EXPOSURE]: The skill explicitly mandates security best practices by requiring PII (Personally Identifiable Information) de-identification for phone numbers and identity details before data is presented to third parties, demonstrating a proactive stance on data privacy.
[COMMAND_EXECUTION]: Mention of text-search operations like 'grep' are utilized as instructional metaphors for the agent to verify its own internal consistency and do not involve shell command execution, system-level modifications, or network-based exfiltration.
[EXTERNAL_DOWNLOADS]: All external references are informational, pointing to reputable scientific and design resources (e.g., ColorBrewer, Adobe Spectrum, and academic research from Cleveland & McGill). No patterns of downloading or executing remote code from untrusted sources were identified.

data-visualization-discipline