The Agent Skills Directory

[Prompt Injection] (SAFE): The instructions follow standard formatting and do not contain any bypass attempts, role-play injections, or commands to ignore previous rules.
[Data Exposure & Exfiltration] (SAFE): No sensitive file paths or hardcoded credentials were detected. References to network listeners are limited to the local loopback address (127.0.0.1) for testing handlers.
[Command Execution] (SAFE): The skill utilizes run_shell_command to execute Go's native testing toolchain (go test -bench and go test -fuzz). This is consistent with the primary purpose of a Go testing expert and does not indicate malicious intent.
[Indirect Prompt Injection] (LOW): The skill operates by running tests on user-provided Go code. This creates a surface where a malicious Go file could include code in a test function or init() block that executes during the testing phase.
Ingestion points: User-provided Go source and test files processed by verify_tests or run_shell_command.
Boundary markers: Absent; the skill does not specify delimiters for the code being tested.
Capability inventory: Includes the ability to execute shell commands and build/verify binaries.
Sanitization: No explicit sanitization or static analysis of the user's Go code is performed before execution.

go-test-expert