The Agent Skills Directory

[NO_CODE]: The skill is entirely declarative, consisting of natural language instructions and a markdown template for the agent to follow. There are no scripts, binaries, or command-line instructions provided.
[SAFE]: Analysis of the skill body and metadata found no evidence of prompt injection, data exfiltration, or obfuscation. The file path specified for output is local and intended for documentation purposes.
[PROMPT_INJECTION]: While the skill processes user-provided context to generate solution variants, it does not possess any tools or capabilities (such as network access or shell execution) that could be exploited if malicious content were present in the input. The attack surface is minimal and limited to text generation.

fpf-variants