The Agent Skills Directory

[PROMPT_INJECTION]: The skill identifies a potential surface for indirect prompt injection as it facilitates the ingestion and rendering of data from external sources (e.g., via DataManager or dataSource) while possessing the capability to perform network operations.
Ingestion points: Data enters the component via the dataSource property or through remote endpoints configured in the <e-data-manager> tag helper as documented in references/data-binding.md.
Boundary markers: The skill does not provide explicit boundary markers or instructions to the agent to disregard embedded commands within the processed data.
Capability inventory: The component is capable of performing network requests to external endpoints for CRUD operations (e.g., url, crudUrl, insertUrl) as described in references/data-binding.md.
Sanitization: The component includes a built-in enableHtmlSanitizer property, enabled by default, which serves to sanitize HTML input and mitigate cross-site scripting (XSS) risks (referenced in references/properties.md).
[EXTERNAL_DOWNLOADS]: The skill documentation includes references to external scripts and stylesheets hosted on the vendor's infrastructure.
Evidence: Mentions of https://cdn.syncfusion.com/ej2/27.1.48/fluent.min.css and https://cdn.syncfusion.com/ej2/27.1.48/dist/ej2.min.js in references/getting-started.md.
[DATA_EXFILTRATION]: The architecture for remote data binding supports transmitting data to external URLs, which could be leveraged as a data exfiltration channel if the configuration or data source is manipulated by an adversary.
Evidence: The use of crudUrl and other endpoint properties in references/data-binding.md enables the component to send card data to external servers.

syncfusion-aspnetcore-kanban