The Agent Skills Directory

[Prompt Injection] (SAFE): The instructions are strictly focused on character analysis and maintain strong grounding requirements ("Strictly based on the provided story text"). There are no attempts to override system safety or extract private instructions.
[Data Exposure & Exfiltration] (SAFE): The skill does not request access to sensitive file paths or credentials. No network tools are authorized, preventing any external data leakage.
[Remote Code Execution] (SAFE): There are no patterns involving external script execution, package installation, or command-line manipulation. The allowed-tools is restricted to Read for internal reference documentation.
[Indirect Prompt Injection] (LOW): The skill processes untrusted user-provided story text. While an adversary could embed malicious instructions within a story, the skill lacks the capabilities (such as shell access, network requests, or file writing) to translate such instructions into a system-level threat.
[Obfuscation] (SAFE): All provided files use clear, readable Markdown and YAML without hidden encoding, zero-width characters, or homoglyph attacks.

character-relationships