The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to use shell commands for project management and auditing, including pnpm install for dependency management, pnpm dev for local development, and rg (ripgrep) for auditing hardcoded color literals.
[EXTERNAL_DOWNLOADS]: The workflow involves cloning a site's source code from a clone_url provided by the platform's list_sites tool and subsequently downloading dependencies via pnpm.
[INDIRECT_PROMPT_INJECTION]: The skill creates a potential attack surface by ingesting external data (brand kits) and user overrides, which are then interpolated into the project's source code files.
Ingestion points: External brand data is fetched using the get_brand_kit tool and through direct user input.
Boundary markers: There are no specific delimiters or boundary markers defined to isolate the ingested brand data from the logic of the code being modified.
Capability inventory: The skill has capabilities to modify the filesystem (app/globals.css, app/layout.tsx) and execute shell commands for building and testing the site.
Sanitization: The skill includes instructions to verify font availability and color contrast, but it does not specify automated sanitization or escaping of the ingested brand strings before they are written to the code files.

apply-branding