The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes standard developer tools including javac, java, and mvn to build and analyze project code. These operations are conducted locally on the user's workspace to verify project health and apply fixes.
[EXTERNAL_DOWNLOADS]: For the remove-deprecated-api pattern, the skill fetches metadata from Maven Central (repo1.maven.org) and Adobe's official documentation (experienceleague.adobe.com). These are well-known, trusted sources used solely to retrieve update information and successor guidance.
[CREDENTIALS_SAFE]: The skill explicitly warns against and provides recipes to remove hardcoded credentials. It promotes secure alternatives like OSGi service users and IMS bearer tokens, consistent with Adobe's security standards for Cloud Service environments.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted code from the user's workspace. It mitigates potential exploitation by using a deterministic Java parser for detection and requiring explicit user confirmation before any edits are applied to the filesystem.
[DATA_EXPOSURE]: The skill is designed to operate locally. It provides clear documentation on its limited network usage, ensuring that sensitive project data remains within the local environment.

code-assessment