The Agent Skills Directory

Indirect Prompt Injection Surface: The skill is designed to ingest and parse user-supplied input via the $ARGUMENTS variable, which is then used to perform actions with the agent's tools. This creates a potential surface where maliciously crafted arguments could attempt to manipulate the agent's logic.
Ingestion point: User input is received through command arguments (SKILL.md).
Capability inventory: The skill utilizes the Bash, Read, Glob, and Grep tools.
Boundary markers: There are no explicit delimiters or instructions provided to separate the untrusted user input from the skill's operational instructions.
Sanitization: The skill does not currently specify steps for validating, filtering, or escaping the content provided in the arguments.
Command Execution Capability: The skill's configuration includes the Bash tool in its allowed-tools list. This enables the agent to execute shell commands to satisfy user requests, which is a powerful capability when combined with the processing of external arguments.

example-command