The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes untrusted Swift source code from the local environment using Glob and Read tools to generate tests. This presents an indirect prompt injection surface where malicious comments in the source code could potentially influence agent behavior, though no specific exploits are present.
[COMMAND_EXECUTION]: The skill utilizes the Bash tool to execute xcodebuild for the purpose of running generated characterization tests. This is a standard operation within the intended scope of Swift development and test automation.
[SAFE]: The skill demonstrates best practices for secret management by recommending the use of mocks for dependencies like network clients and disk access, rather than accessing live sensitive data.

characterization-test-generator