The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill recommends downloading font pairing configurations from https://www.fonttrio.xyz. This domain is an external third-party service and is not recognized as a well-known or trusted provider in this context.
[COMMAND_EXECUTION]: The instructions suggest the command bunx shadcn@latest add https://www.fonttrio.xyz/r/{pairing-name}.json. This command fetches a remote JSON manifest from an external source and uses the shadcn CLI to modify project files and potentially install new dependencies.
[INDIRECT_PROMPT_INJECTION]: The skill analyzes untrusted data from the user's local project files to perform its audit, creating a potential surface for indirect injection attacks.
Ingestion points: Project styling files including *.css, *.scss, globals.css, and tailwind.config.* (SKILL.md).
Boundary markers: Absent; there are no specific delimiters or instructions to ignore embedded commands within the files being audited.
Capability inventory: File system search and read access, combined with the capability to suggest or execute CLI commands like bunx.
Sanitization: Absent; the skill does not specify any validation, escaping, or filtering for the content found in the typography configuration files.

audit-typography