The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill accepts arbitrary natural-language descriptions or branch names via the $ARGUMENTS variable and passes them to a 'general-purpose agent' to execute a workflow pipeline.
Ingestion points: The $ARGUMENTS field in SKILL.md is populated directly by user input.
Boundary markers: There are no visible delimiters or instructions to ignore embedded commands within the provided input.
Capability inventory: The skill allows the use of the Bash(git:*) tool, which indicates the potential for shell command execution within the pipeline.
Sanitization: No sanitization or validation of the $ARGUMENTS content is performed within this skill before it is passed to the next stage of the workflow.

start-feature