The Agent Skills Directory

[SAFE]: The skill content is strictly instructional and provides guidelines for color palette management without incorporating executable code, dangerous system commands, or external dependencies that run automatically.
[INDIRECT_PROMPT_INJECTION]: The skill facilitates the analysis of untrusted data by instructing the agent to audit external source code for color-related literals (ingestion point: project codebase files as described in palette-structure.md). While the instructions do not specify explicit boundary markers to delimit this external content, the skill itself possesses no capabilities for network exfiltration, arbitrary command execution, or automated file writes, and the instructions are scoped to analysis and reporting via a structured format defined in review-output.md (sanitization).

better-colors