The Agent Skills Directory

[SAFE]: The skill implements a robust, multi-phase brainstorming process entirely through prompt instructions and local file operations. It does not attempt to execute arbitrary code, download remote scripts, or bypass safety guidelines.
[COMMAND_EXECUTION]: The skill uses standard system commands (open or xdg-open) specifically to display locally generated HTML visual drafts in the user's browser. This is a limited, functional use of command execution targeting known, agent-generated files within designated paths.
[EXTERNAL_DOWNLOADS]: No external downloads or remote code execution patterns were detected. The skill uses the platform's standard WebSearch tool for research, which is an expected and controlled capability.
[DATA_EXFILTRATION]: There is no evidence of unauthorized network operations or data exfiltration. The skill focuses on processing user-provided topics and saving reports to local paths.
[PROMPT_INJECTION]: The instructions do not contain patterns associated with jailbreaking, ignoring safety filters, or overriding system prompts. The 'self-review' phase actually includes checks for quality and consistency.
[CREDENTIALS_UNSAFE]: No hardcoded secrets, API keys, or instructions to access sensitive configuration files (like .env or SSH keys) were found.

brainstorm