The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted user data (Commit type, Ticket ID, and Short title) and interpolates these values directly into shell commands. This creates a vulnerability surface for command injection if an attacker provides inputs containing shell metacharacters.
Ingestion points: SKILL.md (Gather inputs section where user provides type, ID, and title).
Boundary markers: None. The inputs are directly used to construct variables for the shell.
Capability inventory: Shell command execution via git, grep, and echo (including file append operations on .gitignore).
Sanitization: The skill provides a sanitization rule ('Slugify the title: lowercase, replace spaces/underscores with hyphens, strip special chars'), which reduces the risk for the title field. However, similar sanitization is not explicitly mandated for the 'Commit type' or 'Ticket ID' fields.
[COMMAND_EXECUTION]: The skill executes multiple shell commands to manage a git environment. This includes git worktree add, which creates new directories and branches, and echo >> .gitignore, which modifies project configuration files. While these are the intended functions of the skill, they represent a significant capability tier.

create-worktree