The Agent Skills Directory

[SAFE]: The skill is a utility for maintaining design systems through static analysis. It uses restricted shell commands to identify hardcoded style values and verify token architecture consistency.
[INDIRECT_PROMPT_INJECTION]: The skill possesses a data ingestion surface as it processes untrusted codebase files, but this is inherent to its primary purpose as an auditing tool. The risk is considered safe as the logic is centered on regex-based pattern matching rather than instruction execution.
Ingestion points: Source code files across various formats (CSS, SCSS, TS, JSX, Vue, Twig), project configuration files (.ds-ops-config.yml), and external design data via Figma MCP.
Boundary markers: No explicit delimiters are specified for separating untrusted code content from agent instructions, though the skill's logic focuses on specific property-value pairs.
Capability inventory: The agent is granted limited permissions: file reading/writing and restricted bash commands (rg, grep, git, find). It lacks general network access or arbitrary code execution capabilities.
Sanitization: The skill utilizes regex patterns (e.g., for hex colors and pixel values) which naturally isolates the data of interest from potential surrounding malicious instructions.

token-compliance