The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to analyze project source files (HTML, JavaScript, Vue, React, CSS) to identify unused styles. This ingestion of data from the user's workspace represents a potential attack surface.
Ingestion points: The ComponentBasedPurge class in references/rule.md scans files in the component directory. The DynamicPurgeCSS class uses Puppeteer to navigate to local or remote URLs to track CSS usage and interactions.
Boundary markers: There are no explicit delimiters or instructions to the agent to disregard potential instructions embedded within the source files being analyzed.
Capability inventory: The skill demonstrates capabilities for writing files to disk (fs.writeFileSync), scanning local directories, and launching a browser process (puppeteer.launch) to interact with web content.
Sanitization: No specific sanitization or filtering is applied to the content of the files read during the analysis phase to prevent the interpretation of embedded instructions.
[EXTERNAL_DOWNLOADS]: The documentation references and provides installation commands for several third-party Node.js packages and GitHub Actions.
Evidence: Mentions standard tools such as purgecss, uncss, puppeteer, and official GitHub Actions like actions/checkout@v4 and actions/setup-node@v4.
[COMMAND_EXECUTION]: The skill provides examples of executing CLI commands to perform CSS cleanup and manage dependencies.
Evidence: Usage examples include npx purgecss --css styles.css --content index.html --output build/, uncss index.html > cleaned.css, and various npm install commands.
[DYNAMIC_EXECUTION]: The skill includes scripts that programmatically interact with the file system and external environments to process and optimize project assets.
Evidence: Scripts like CSSPerformanceMonitor and ComponentBasedPurge use the Node.js fs and path modules to read, analyze, and write project files at runtime.

unused-css