hermes-agent

SUSPICIOUS. The skill is largely coherent with its stated purpose as a guide for a powerful autonomous agent, and the main Hermes installer appears same-org and officially documented. Risk is elevated by broad agent capabilities, third-party CLI usage, scan-bypass guidance for ClawHub, and especially transitive installation of external skills from URLs/GitHub repos, which can inherit extensive local access.

No malicious code is evident because this fragment is documentation only. The primary finding is a serious credential exposure: a plaintext ElevenLabs API key value is embedded directly in the content. This can enable unauthorized third-party usage/billing abuse and account compromise if the key is still valid. Immediate key revocation/rotation and removal of the secret from all published artifacts/history is recommended.