The Agent Skills Directory

[SAFE]: The skill instructions and documentation focus entirely on corporate branding standards, including visual identity, typography, and tone of voice. There are no patterns suggesting prompt injection, behavioral overrides, or safety bypasses.
[SAFE]: Data processing is limited to formatting configurations for documents. The provided Python scripts (apply_brand.py and validate_brand.py) perform logic-based transformations and regex-based validation without sensitive file access, network operations, or hardcoded credentials.
[SAFE]: No obfuscation techniques were detected. All code and documentation are in plain text, and hex codes for colors are standard and verified.
[SAFE]: External dependencies are not used. The scripts rely solely on Python's standard library (e.g., re, json, dataclasses). There is no remote code execution or dynamic execution of untrusted code.
[SAFE]: Indirect prompt injection risk is negligible. While the BrandValidator processes text content, it lacks dangerous 'sinks'—it does not write to the file system, execute shell commands, or make network requests based on the input data.
[SAFE]: No persistence mechanisms, privilege escalation attempts, or dynamic context injections (!command syntax) were found in the skill metadata or body.

applying-brand-guidelines