The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes standard Maven (./mvnw) and Gradle (./gradlew) wrappers to execute build, test, and verification tasks within the project environment.
[EXTERNAL_DOWNLOADS]: The provided templates reference well-known dependencies and tool distributions from trusted public repositories such as Maven Central.
[SAFE]: The skill incorporates robust security patterns, including OAuth2/JWT implementation, environment-based configuration for secrets, and extensive testing with TestContainers. No signs of obfuscation, exfiltration, or persistence were found.
[SAFE]: A standard indirect prompt injection surface is present due to the processing of user-provided project files.
Ingestion points: Project source code and build configuration files read by the agent.
Boundary markers: None; project data is treated as context for architectural tasks.
Capability inventory: Shell command execution via build tool wrappers (SKILL.md).
Sanitization: None, as the skill is designed to interact with and verify development source code.

java-architect