The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the vendor's dxs CLI and the jq utility to perform configuration management tasks. These tools are used as intended within the platform's development lifecycle for fetching, validating, and updating JSON configuration files.
[REMOTE_CODE_EXECUTION]: The skill assists in generating TypeScript code for component lifecycle hooks (on_init). This code is executed by the Datex Studio runtime to compute component properties. The skill provides standard, safe templates for these scripts as part of the component authoring process.
[PROMPT_INJECTION]: The skill defines a component architecture that ingests external data (such as HTML strings) via inParams. This represents a potential surface for indirect prompt injection if untrusted data is processed.
Ingestion points: External content is passed into the component via parameters such as inParams.html or inParams.marker.
Boundary markers: The skill does not explicitly instruct the agent to use specific delimiters or boundary markers when handling or interpolating this data during the creation or preview process.
Capability inventory: Components have the capability to render content in an iframe and execute localized startup scripts. They also have access to the $shell global for UI interactions like opening dialogs.
Sanitization: The skill mitigates risks by recommending the use of encodeURIComponent for building data URIs and highlights the application of platform-level security measures including the safeUrl pipe and Content Security Policy (CSP).

embed-creator