The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests user raw input into the agent's execution context via the $ARGUMENTS variable.
Ingestion points: User-supplied input is captured through the argument-hint field and passed into the $ARGUMENTS variable within SKILL.md.
Boundary markers: No delimiters, quoting, or instructions to treat input as untrusted data are provided to prevent the agent from following instructions embedded within the arguments.
Capability inventory: The skill allows execution of git commands via Bash(git:*) and has local Read and Write file system permissions.
Sanitization: There is no evidence of filtering, escaping, or schema validation for the $ARGUMENTS before they are passed to the secondary workflow pipeline.

start-release