The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill uses npx @salesforce-ux/slds-linter@latest, which fetches the latest version of the linter from the official npm registry. This is a standard method for accessing Salesforce's development tools. \n- [REMOTE_CODE_EXECUTION]: The linter package downloaded via npx is executed locally to perform the audit. Execution is limited to the official @salesforce-ux/slds-linter tool. \n- [COMMAND_EXECUTION]: The skill runs shell commands to invoke the SLDS linter and a local Node.js script (scripts/analyze-quality.cjs). These commands are used as intended for static code analysis. \n- [PROMPT_INJECTION]: The skill processes user-provided source code (CSS, HTML, and JS) to generate its reports, which presents an indirect prompt injection surface. \n
Ingestion points: Source code from Lightning Web Components is read by the analysis script and the linter (SKILL.md). \n
Boundary markers: The process does not explicitly define delimiters for the ingested code snippets within the analysis script's output, though the final report template uses markdown sections. \n
Capability inventory: The skill's capabilities are limited to reading local files and generating text-based summaries. It does not have access to sensitive credentials or external network services for data exfiltration. \n
Sanitization: The analysis script uses regular expressions to extract specific patterns rather than executing the code, reducing the risk of malicious instructions being interpreted as commands.

validating-slds