The Agent Skills Directory

[COMMAND_EXECUTION]: The skill makes extensive use of the npx cypress tap command to manage testing sessions. This includes commands for dispatching test specs (run), monitoring session state (status), and retrieving detailed command logs (reporter, command). These operations are consistent with the skill's primary purpose of test automation and debugging.
[INDIRECT_PROMPT_INJECTION]: The skill provides commands (dom, aria, inspect) that allow the agent to ingest the live HTML, accessibility tree, and computed styles of the application under test. This data is untrusted as it originates from the application environment and could potentially contain malicious instructions.
Ingestion points: Data enters the agent's context through the output of dom, aria, and inspect commands as described in references/reading-the-app.md.
Boundary markers: The instructions do not prescribe specific delimiters or "ignore instructions" prompts when processing the content returned from these readers.
Capability inventory: The agent possesses shell execution capabilities via npx and file-write capabilities (e.g., redirecting JSON output to files as demonstrated in references/reading-results.md).
Sanitization: There is no mention of sanitizing, escaping, or validating the ingested DOM content before the agent analyzes it.
[EXTERNAL_DOWNLOADS]: The skill relies on npx for executing Cypress-related tooling. This may involve downloading the Cypress binary or CLI extensions from the NPM registry if they are not already cached. These downloads originate from a well-known service (NPM) and target the official Cypress toolset, which is consistent with standard Node.js development practices.

cypress-tap