The Agent Skills Directory

[DATA_EXFILTRATION]: The skill instructions require the agent to access and read the .env file to retrieve sensitive credentials including TEST_EMAIL, ADMIN_EMAIL, TEST_PASSWORD, and ADMIN_PASS. While this access is essential for the skill's purpose of automated authentication, it involves the exposure of sensitive local files.
[PROMPT_INJECTION]: The skill creates a surface for indirect prompt injection by navigating to and interacting with an external web application (APP_URL). Maliciously crafted content on the target site could theoretically attempt to influence the agent's actions during the login sequence.
Ingestion points: External web application via the APP_URL variable.
Boundary markers: No specific delimiters or "ignore instructions" warnings are defined for the browser interaction steps.
Capability inventory: Playwright browser automation tools (navigate, click, fill, wait).
Sanitization: No sanitization or content validation of the target page is performed prior to interaction.
[COMMAND_EXECUTION]: The skill utilizes Playwright browser automation tools to execute navigational and interactive commands (e.g., clicking buttons and filling text boxes) based on data retrieved from environment variables.

auth-handler