The Agent Skills Directory

[SAFE]: The skill utilizes the AskUserQuestion tool for all user interactions, ensuring a controlled and structured communication flow.
[SAFE]: Data persistence is limited to the local thoughts/ directory, which is the expected location for user-specific brainstorm documents and does not involve unauthorized file system access.
[SAFE]: No external network operations, hardcoded credentials, or obfuscated code patterns were detected.
[SAFE]: The skill integrates with internal platform capabilities (like the file-review plugin) and follows standard handoff procedures to other development commands without bypassing safety constraints.
[PROMPT_INJECTION]: While the skill processes user-provided ideas, it uses Markdown structural boundaries (headers and Q&A blocks) to distinguish between instructions and user-provided content, minimizing the risk of instruction override.

brainstorming