The Agent Skills Directory

[SAFE]: The skill defines operational guidelines for a brainstorming workflow. It specifies trigger phrases, plan mode state management, and file placement conventions for design specifications within the project's documentation directory.
[INDIRECT_PROMPT_INJECTION]: The skill establishes an attack surface by instructing the agent to trigger specific behaviors based on user-provided natural language phrases. However, the resulting actions are confined to internal tool calls (e.g., AskUserQuestion, ExitPlanMode) and structured file writes to project-specific paths (docs/superpowers/specs/). There is no evidence of instructions that would allow user input to bypass safety filters or execute arbitrary commands.
[COMMAND_EXECUTION]: The instructions involve file system operations such as writing and deleting temporary design specifications. These operations are restricted to a defined subdirectory and are part of the intended document lifecycle management, presenting no risk of unauthorized file access or privilege escalation.

brainstorming-conventions