The Agent Skills Directory

[COMMAND_EXECUTION]: The skill invokes shell commands including npm install for dependency management and sf project retrieve for fetching Salesforce Content Type metadata. It also uses npx tsc for type-checking generated code.
[EXTERNAL_DOWNLOADS]: The skill downloads the @salesforce/ui-bundle-template-feature-cms-toolkit package from the official npm registry. As this is a well-known package from a trusted organization, it is considered safe.
[INDIRECT_PROMPT_INJECTION]: The skill processes content fetched from Salesforce CMS, which represents a potential indirect injection surface when rendered into the UI.
Ingestion points: CMS data is ingested via toolkit functions getCmsContentByUrl and getCmsContentByKey (found in assets/react/useCmsItem.ts and assets/angular/cms-item.service.ts).
Boundary markers: The skill explicitly instructs the agent and the developer to use sanitizers, such as Angular's built-in DomSanitizer or the DOMPurify plug in React (described in references/failure-modes.md and references/heuristic-render-rules.md).
Capability inventory: The skill possesses the capability to execute shell commands (npm, sf, npx) and write generated React/Angular components and TypeScript definitions to the project directory.
Sanitization: RichText content is decoded using the decodeRichHtmlEntities utility and is subjected to framework-level sanitization (e.g., [innerHTML] in Angular) before being injected into the DOM.

experience-cms-content-render