The Agent Skills Directory

[SAFE]: The skill provides a structured way to manage multiple independent tasks. It uses platform tools like Read, Write, and Bash to interact with project-specific state files and templates, which is consistent with its stated purpose as a workflow manager.
[INDIRECT_PROMPT_INJECTION]: The skill ingests user-provided feature names which are stored in state files and used to generate file paths. While this is a common pattern for CLI-like tools, it represents a standard injection surface.
Ingestion points: Feature names provided in the plan command (SKILL.md).
Boundary markers: The skill uses AskUserQuestion to confirm plans before execution, providing a human-in-the-loop check.
Capability inventory: Uses Write to create files and Bash for internal logic (SKILL.md).
Sanitization: No explicit string sanitization is described for feature names, though the system uses standard templates.

pdca-batch