The Agent Skills Directory

[Prompt Injection] (SAFE): No instructional patterns were found that attempt to override the AI agent's core safety guidelines or behavior. All instructions are context-specific to book writing.
[Data Exposure & Exfiltration] (SAFE): No patterns for accessing sensitive files (like .ssh or .aws) or network requests (curl, wget) were detected.
[Obfuscation] (SAFE): All text is plain markdown; no Base64, zero-width characters, or hex encoding was used to hide content.
[Remote Code Execution] (SAFE): The skill lacks any programming scripts (.py, .js, .sh) or dependency files (package.json, requirements.txt) that could trigger remote code execution.
[Persistence] (SAFE): No commands or instructions were found that attempt to modify shell profiles or system startup tasks.
[Dynamic Execution] (SAFE): There is no logic for runtime code compilation, eval calls, or dynamic library loading.

narrative-nonfiction