The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the coval CLI to perform administrative tasks such as listing agents, retrieving agent details, and creating test sets and cases. These commands are legitimate operations within the context of the Coval platform.
[EXTERNAL_DOWNLOADS]: The skill makes network requests to https://api.coval.dev using curl to create test cases programmatically. As these requests target the official API of the skill's author (coval-ai), they are documented as safe vendor-related operations.
[CREDENTIALS_UNSAFE]: Authentication is handled securely. The skill guides users to use coval login or the $COVAL_API_KEY environment variable, avoiding the use of hardcoded secrets and following industry standards for secret management.
[PROMPT_INJECTION]: The skill processes user-provided scenario text but implements safety guidance by recommending single quotes in shell commands to prevent unintended shell interpolation of special characters within test case inputs.

build-test-suite