The Agent Skills Directory

[DYNAMIC_EXECUTION]: The skill implements a local Java-based analysis engine. The scripts/analyze.sh script compiles the analyzer's source code (located in the analyzer/ directory) into a temporary cache at runtime. This allows for high-performance, deterministic AST-level scanning of the user's workspace without relying on external services.- [EXTERNAL_DOWNLOADS]: The remove-deprecated-api pattern utilizes scripts/detect.sh to fetch metadata from Maven Central (repo1.maven.org) and Adobe Experience League (experienceleague.adobe.com). These calls are used to resolve the latest AEM SDK and analyzer plugin versions to ensure the rule set is current. Both sources are well-known and trusted.- [COMMAND_EXECUTION]: The skill performs local command execution of mvn (Maven) to compile the project and run specialized AEM analysis plugins. This is necessary for verifying that code changes are correct and buildable within the AEM environment.- [INDIRECT_PROMPT_INJECTION]: The skill has an indirect prompt injection surface as it ingests and processes source code from the local workspace. However, the risk is low as the analyzer uses a structured AST parser (javac) for inspection only, and the risk is mitigated by the local execution environment and specific hardening measures, such as disabling external entities in its XML parser configuration (PomUnit.java), to prevent exploitation.

code-assessment