The Agent Skills Directory

[COMMAND_EXECUTION]: The skill relies on the execution of shell commands through the rota CLI tool to create (rota item create), audit (rota item shipped), and remove (rota item rm) backlog items. These commands are executed directly in the shell environment using data provided by the user.
[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it processes untrusted user input (e.g., brain-dumps of bugs, tasks, or features) and interpolates this content directly into shell command arguments (e.g., --title, --desc). If the agent does not properly escape shell metacharacters or quotes within the user-supplied text, it could lead to command injection.
Ingestion points: User-provided text describing work items in SKILL.md (Step 2) and milestone specs in milestone-spec-audit.md (Step 2.5).
Boundary markers: Absent. The instructions do not provide guidance on how to delimit or escape user input when generating shell commands.
Capability inventory: The skill utilizes shell capabilities for item management, including creation, deletion, and modifying issue labels through the rota CLI and piped commands like jq.
Sanitization: The skill lacks explicit instructions or logic to escape, validate, or filter user-supplied content before it is interpolated into commands.

rota-capture