The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the Bash tool to execute curl for interacting with the Discord API. The request body is dynamically generated from user-provided $ARGUMENTS. This pattern poses a risk of command injection if the agent does not properly escape or sanitize these arguments when constructing the final shell command.
[DATA_EXFILTRATION]: The skill transmits information to a webhook_url. While the intended destination is Discord (a well-known service), the skill's logic allows sending data to any URL stored in config.json or provided by the user during the setup phase.
[PROMPT_INJECTION]: The skill processes untrusted user input to determine message content and influence command parameters, creating a surface for indirect prompt injection.
Ingestion points: Untrusted data enters the agent context via the $ARGUMENTS variable in SKILL.md.
Boundary markers: No explicit delimiters or "ignore" instructions are present to separate untrusted user data from the core command logic.
Capability inventory: The skill possesses the capability to execute shell commands (Bash) and perform network operations (curl).
Sanitization: The instructions do not specify any sanitization or escaping protocols for the input before it is interpolated into the shell command or JSON payload.

notify-discord