The Agent Skills Directory

[SAFE]: The skill implements a persona focused on value investing and business analysis. The instructions in SKILL.md and SKILL.v2.md use strong rhetorical language ('shamanic possession', 'wealth judgment') to maintain a consistent character, but these do not contain instructions to bypass safety filters or exfiltrate data.
[SAFE]: The 'System Prompt' section in SKILL.v2.md defines a structured five-step method (Filter, Analyze, Evaluate, Decide, Hold) for problem-solving, which is a benign instructional framework.
[SAFE]: All identified parameters (temperature, top_p) and metadata fields are standard configuration for LLM behavior and do not introduce security risks.

shaman-buffett