Hermes Agent is in use, and it feels quite smooth so far: - One-click import of OpenClaw configurations and memories from the same server (though I haven't done this) - Asked it to fix a Telegram communication bug; after resolving it, it automatically created a related Skill - The experience with the GPT-5.4 model actually feels better within Hermes than in OpenClaw... - Some details are well done, such as the token burn progress and execution details. Hurry up and surpass the Claude Code experience... I'll explore it thoroughly, hoping to discover more pleasant surprises.