The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill includes code examples, such as the 'exportWithRemoteTemplate' method in 'references/scenarios/business-scenarios.md', that fetch Excel templates from arbitrary remote URLs using 'new URL(tplUrl).openStream()'. This pattern creates a significant Server-Side Request Forgery (SSRF) risk if the URL is influenced by untrusted input.
[COMMAND_EXECUTION]: Several code snippets demonstrate writing generated Excel files to local file system paths provided via variables, such as 'EasyExcel.write(outputPath)'. The lack of path sanitization in these examples makes the implementation susceptible to path traversal attacks, potentially allowing an attacker to overwrite critical system files.
[INDIRECT_PROMPT_INJECTION]: The skill processes external data (DTOs, Maps) and template files (.xlsx), which represents a vulnerability surface for indirect prompt injection. Malicious instructions could be hidden in processed data or template metadata. Mandatory Evidence Chain: 1. Ingestion points: Data from DTO/Map objects and templates loaded from local paths or remote URLs. 2. Boundary markers: No delimiters or warnings to ignore instructions within the template are implemented. 3. Capability inventory: File system writing via 'EasyExcel.write()' and network retrieval of templates. 4. Sanitization: The provided examples do not include logic for sanitizing data values or template contents before processing.

easyexcel-fill