The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses standard development commands for environment discovery and project analysis.
It uses npx cypress --version to determine the project's Cypress version if it cannot be found in package.json.
It prefers using grep to search for specific content within files rather than reading entire files, which is an efficient and safe practice for large projects.
[CREDENTIALS_UNSAFE]: The skill explicitly instructs the agent to follow security best practices by using cy.env() for credentials, passwords, or secrets, rather than hardcoding them in tests.
[EXTERNAL_DOWNLOADS]: The skill references official, trusted documentation from docs.cypress.io.
[INDIRECT_PROMPT_INJECTION]: The skill processes user-provided instructions to generate or fix code. It includes guardrails such as requiring the agent to identify the task type and parameters before execution and explicitly warns against running tests unless requested by the user.

cypress-author