The Agent Skills Directory

[COMMAND_EXECUTION]: Uses the open command in a bash script to launch the system browser to the Zyte agent setup URL.
[REMOTE_CODE_EXECUTION]: Executes local scripts using source (for text files) or powershell -File (for PowerShell scripts) based on a file path provided by the user. This dynamic execution of scripts from user-supplied paths is a risk if the user is social-engineered into providing a path to a malicious file.
[COMMAND_EXECUTION]: Constructs shell commands to write a user-provided Project ID to a local file. The direct interpolation of user input into a shell command (echo "PROJECT_ID" > ...) without explicit sanitization creates a potential command injection surface if the input contains shell metacharacters.
[EXTERNAL_DOWNLOADS]: Instructs the user to download a configuration file from https://app.zyte.com/agent-setup. This download targets a well-known service associated with the skill's primary function.
[DATA_EXFILTRATION]: Accesses sensitive environment variables ZYTE_API_KEY and SHUB_APIKEY. The skill includes explicit instructions to prevent the agent from displaying these values in the conversation transcript, which is a recognized security best practice for credential handling.

scrape-zyte-login