The Agent Skills Directory

[SAFE]: No security issues were detected. The skill uses standard file system tools (grep, read, bash) to audit project source code and configuration files for design token consistency. It adheres to best practices by verifying reference paths and configuration settings before proceeding.
[INDIRECT_PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface as it is designed to ingest and analyze external project source code and token files. This risk is inherent to its primary purpose as an auditing tool and is mitigated by the skill's focused analysis workflow.
Ingestion points: The agent reads project files such as .resolver.json, CSS custom property definitions, SCSS variable maps, and JavaScript theme objects.
Boundary markers: The instructions do not define explicit delimiters for the content read from external files.
Capability inventory: The skill uses Read, Write, and Bash (cat, find, grep, rg) to process files and generate reports.
Sanitization: No specific sanitization of external content is described prior to analysis.

theme-audit