The Agent Skills Directory

[SAFE]: The skill is purely informational/advisory. It analyzes text-based marketing profiles and generates a set of copy recommendations and layout specifications for the user to manually apply to their social media accounts.
[DATA_EXPOSURE_EXFILTRATION]: No network operations (curl, wget, fetch) or hardcoded credentials were detected. The skill only reads local markdown files containing brand and audience information to inform its recommendations.
[REMOTE_CODE_EXECUTION]: There are no commands for package installation (npm, pip) or remote script execution. The skill references other internal tools like 'ideogram.md' and 'nano-banana.md' for image generation, which is a standard pattern for tool chaining within the platform.
[COMMAND_EXECUTION]: No shell commands, subprocess calls, or privileged operations are present in any of the skill's files.
[INDIRECT_PROMPT_INJECTION]: While the skill ingests external brand data, it has no exploitable capabilities (no file writing, no network access, no code execution), making the risk of indirect injection negligible.
[OBFUSCATION]: Analysis of the text and metadata revealed no Base64 encoding, zero-width characters, or homoglyph attacks.

profile-optimization