The Agent Skills Directory

[Prompt Injection] (SAFE): The skill does not contain instructions to override system prompts or bypass safety filters.
[Data Exposure & Exfiltration] (SAFE): The skill suggests writing test logs to /tmp/test-output.txt. While this is a shared directory on multi-user systems, it is a standard practice for temporary data and does not involve network exfiltration or access to sensitive credential files.
[Remote Code Execution] (SAFE): No remote scripts are downloaded or executed. The skill refers to local test runners (npm, pytest) which is consistent with its stated purpose.
[Indirect Prompt Injection] (LOW): The skill reads data from test outputs which could theoretically contain malicious instructions (Category 8).
Ingestion points: /tmp/test-output.txt via tail, wc, and grep.
Boundary markers: None specified in the redirection command.
Capability inventory: Local command execution (test runners) and file system read access.
Sanitization: None. However, this is a standard risk for any tool processing external command output and is acceptable for the skill's primary purpose.

running-tests