The Agent Skills Directory

[SAFE]: The skill defines a brainstorming and questioning workflow aimed at helping users refine plans and decisions. It contains no instructions that attempt to bypass safety guidelines, exfiltrate data, or execute unauthorized code.
[INDIRECT_PROMPT_INJECTION]: The instructions direct the agent to explore the local environment (filesystem and tools) to find facts instead of asking the user. This creates an ingestion surface for untrusted data from the filesystem which could theoretically influence the agent's behavior; however, the skill's narrow focus on questioning the user and requiring confirmation for decisions minimizes the risk.

grilling