The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes untrusted data via the $ARGUMENTS variable and the --do dispatcher, creating a surface for indirect prompt injection.
Ingestion points: User-supplied input is directly interpolated into the execution logic within SKILL.md via the $ARGUMENTS variable.
Boundary markers: The skill does not use delimiters (such as XML tags or specific quotation markers) to isolate the user input from the instructional context.
Capability inventory: The skill possesses high-privilege capabilities including the Bash tool, SlashCommand execution, and broad file system access (Read, Grep, Glob).
Sanitization: There are no instructions provided to validate, sanitize, or escape the user intent before it is processed by the dispatcher to trigger other workflows.
[COMMAND_EXECUTION]: The skill utilizes the Bash tool to perform operations and incorporates a looping mechanism (--next --auto) that chains workflow steps automatically. This automation increases the potential impact if the agent is misled by malicious input provided via the dispatcher.

gsd-progress