Closed vs open

Claude+Desktop vs Hermes+OpenClaw

Run locally

ollama pull hermes3
client = OpenAI(base_url="http://localhost:11434/v1", api_key="ollama")
client.chat.completions.create(model="hermes3", tools=[...])

Tool-call demo

Same 5 prompts: base Llama vs Hermes

Count valid JSON tool_calls — Hermes wins

Agent OS

Gateway · Heartbeat · Telegram/Discord

Demo sequence

1. ollama pull hermes3
2. openclaw init
3. openclaw skills install …
4. Connect Telegram · openclaw start

Example YAML

model:
  base_url: http://127.0.0.1:11434/v1
  name: hermes3
channels:
  telegram: { enabled: true }

Skill package

skill.yaml + prompt + tools · ClawHub

End-to-end flow

Telegram → Hermes plan → skills → reply

Heartbeat = 8am daily brief without user prompt

LangGraph vs OpenClaw

OpenClaw + Hermes

1. Ollama Hermes + 5 tool prompts
2. Point OpenClaw to local endpoint
3. Install 1 ClawHub skill
4. Compare tool JSON validity vs base Llama