The Agent Skills Directory

[PROMPT_INJECTION]: The skill is vulnerable to command injection through user-controlled data.
Ingestion points: User input for the branch name collected in Step 1.
Boundary markers: Absent; the branch name is directly interpolated into the shell command string in Step 3.
Capability inventory: Execution of shell commands via the Bash tool, including git fetch and git worktree add.
Sanitization: Absent; the skill provides no instructions to validate, escape, or sanitize the user-provided branch name before it is used in a command. This allows a user to terminate the intended command and execute arbitrary bash commands.
[COMMAND_EXECUTION]: The skill utilizes the Bash tool to perform file system and git operations. It specifically instructs the agent to create directories outside the current working directory (../), which increases the risk surface when combined with the lack of input validation.

feature-branch