The Agent Skills Directory

[COMMAND_EXECUTION]: The login.js script contains an openBrowser function that uses child_process.exec to launch a web browser. The command is constructed using a URL that incorporates a base URL derived from an environment variable (WALKINTO_URL). If this environment variable is manipulated to include shell metacharacters, it could lead to arbitrary command execution on the host system.- [CREDENTIALS_UNSAFE]: The skill manages authentication by reading from and writing to a local token file located at ~/.config/walkinto/token. While this is a standard mechanism for session persistence in CLI tools, accessing and managing sensitive credential files on the local filesystem is a security-sensitive operation.- [PROMPT_INJECTION]: The skill processes user-generated content from external WalkInto helpdesk tickets via the tickets.js script. This represents an indirect prompt injection surface where malicious instructions embedded in a ticket could attempt to influence the agent's behavior during ticket triaging or reply drafting.
Ingestion points: The view command in tickets.js retrieves and displays the subject, description, and full message thread of helpdesk tickets from the remote API.
Boundary markers: None identified. The scripts output retrieved ticket data directly to the console without delimiters or instructions to ignore embedded commands.
Capability inventory: The skill possesses significant capabilities, including the ability to reply to customer tickets, update ticket statuses/priority, and execute local shell commands via the provided Node.js scripts.
Sanitization: No sanitization or filtering of the ticket content is performed before it is presented to the agent.

walkinto