evolve

SUSPICIOUS: the skill’s core purpose is coherent, but it instructs the agent to install/load other skills from personal GitHub marketplaces and to ingest untrusted web/GitHub content while generating executable guidance and publishing outputs. The main issues are transitive skill trust and indirect prompt-injection exposure, not confirmed malware.

SUSPICIOUS: The skill’s broad autonomous behavior is consistent with its stated self-evolving purpose, but that purpose is itself high-scope. The main risk is autonomy plus unspecified version-check/auto-update behavior whose trust path is not visible here; there is no clear evidence of credential theft, third-party interception, or confirmed malware in the provided excerpt.