The Agent Skills Directory

[COMMAND_EXECUTION]: The execute function in run.py is vulnerable to command injection. It constructs a shell command string by interpolating the package and timeout variables directly from the input args dictionary without validation. This command is then executed via subprocess.run(cmd, shell=True). An attacker could inject arbitrary shell commands by providing malicious values for these parameters.
Evidence: cmd = f"CGO_ENABLED=0 go test {package} -v -json -timeout {timeout}" and subprocess.run(cmd, shell=True, ...) in run.py.
[REMOTE_CODE_EXECUTION]: The lack of input sanitization in the command execution logic allows for arbitrary code execution on the host system if the agent is manipulated into passing malicious arguments.
[PROMPT_INJECTION]: The skill presents an indirect prompt injection surface by ingesting output from an external tool (go test) and using it to generate an analysis_prompt for the agent's next steps.
Ingestion points: The skill parses stdout from the go test command in run.py.
Boundary markers: No boundary markers or instructions to ignore embedded commands are present in the result processing.
Capability inventory: The skill can execute shell commands via subprocess.run in run.py.
Sanitization: No sanitization or escaping is performed on the test output before it is incorporated into the skill's response.

self-test