The Agent Skills Directory

[SAFE]: No executable code or scripts.
The skill is strictly limited to conversational brainstorming and does not contain any scripts, binaries, or automated tasks.
The Boundaries section explicitly prohibits the agent from writing code.
[SAFE]: No network or file system access.
The analysis found no network commands (e.g., curl, wget) or file system manipulation tools.
The skill does not request or use any sensitive API keys or credentials.
[SAFE]: Secure input handling.
The RECEIVE workflow phase mandates that the agent summarizes and confirms the user's idea before proceeding, which helps mitigate accidental obedience to malicious instructions embedded in user data.
[SAFE]: No external dependencies or obfuscation.
All logic is defined in plain-text markdown files without the use of third-party packages (npm/pip) or obfuscated content (Base64/Hex).

riff