The Agent Skills Directory

Indirect Prompt Injection (LOW): The skill accepts arbitrary natural language input to generate task cards. While it lacks explicit sanitization, the risk is limited to the generation of low-quality or misleading task documentation. Ingestion points: User task descriptions via LLM interaction. Boundary markers: None. Capability inventory: Agent-side file writing to src/renderer/TODO.md and local execution of the benign breakdown_validator.py script. Sanitization: None.
COMMAND_EXECUTION (INFO): The provided Python script scripts/breakdown_validator.py performs static analysis on text files. It does not execute arbitrary shell commands or access the network.

task-breakdown