The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests Java source code and test files from the workspace, creating an attack surface where instructions embedded in project data could influence the agent.
Ingestion points: Source and test files in src/test/java and build configuration files (Maven/Gradle) are processed.
Boundary markers: The skill lacks explicit instructions or markers to distinguish between the agent's core guidelines and potentially malicious content within the ingested files.
Capability inventory: The skill uses shell commands like mvn test and gradle test to execute the code, providing a capability that could be triggered by payloads in the source files.
Sanitization: No validation or sanitization is performed on the ingested code prior to execution.
[COMMAND_EXECUTION]: The skill instructs the agent to execute standard shell commands (mvn test, gradle test) to perform local validation of Java unit tests.
[EXTERNAL_DOWNLOADS]: The changelog mentions that the skill was originally imported from a repository on a well-known service (github.com/github/awesome-copilot).

java-junit