The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses shell commands to manage files and integrate with version control systems as part of its planning workflow.
Executes git status, git diff, and git mv to detect user modifications and move refined task files to the appropriate directories.
Executes a vendor-provided initialization script create-folders.sh located in the plugin's root directory to set up the necessary project structure.
[INDIRECT_PROMPT_INJECTION]: The skill processes external task specifications and user-supplied arguments which could theoretically contain instructions intended to influence sub-agent behavior.
Ingestion points: Reads content from draft task files located in .specs/tasks/draft/ and parses user input from $ARGUMENTS.
Boundary markers: Uses clear headers like Task File: and Task Title: when passing content to sub-agents, though it does not explicitly instruct agents to ignore embedded directives in the draft content.
Capability inventory: The orchestration involves launching multiple sub-agents (sdd:researcher, sdd:software-architect, etc.), writing to the file system, and performing Git operations.
Sanitization: No explicit sanitization or escaping of the ingested task file content is performed before it is interpolated into the prompts for sub-agents.

plan-task