Deploying Hermes Agent in Containers

1. Starting the Container

Set the image

1
export IMAGE=nousresearch/hermes-agent:v2026.6.5

Set directory permissions

1
2
mkdir -p hermes-agent-home
chmod -R 777 hermes-agent-home

Set environment variables

1
2
3
cat <<EOF > hermes-agent-home/.env
GATEWAY_ALLOW_ALL_USERS=true
EOF

Start hermes-agent

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
nerdctl run -d \
  --name hermes-agent \
  --restart always \
  --security-opt apparmor=unconfined \
  --security-opt seccomp=unconfined \
  --user hermes \
  --network host \
  -v $(pwd)/hermes-agent-home:/opt/data \
  -v $(pwd)/openclaw-home/.openclaw:/opt/data/.openclaw \
  $IMAGE gateway run

Start the hermes-agent dashboard

1
2
3
4
5
6
7
8
9
nerdctl run -d \
  --name hermes-dashboard \
  --restart always \
  --security-opt apparmor=unconfined \
  --security-opt seccomp=unconfined \
  -p 9119:9119 \
  -v $(pwd)/hermes-agent-home:/opt/data \
  -e GATEWAY_HEALTH_URL=http://${node_IP}:8642 \
  $IMAGE dashboard --host 0.0.0.0 --insecure

Remove the containers

1
nerdctl rm hermes-agent hermes-dashboard --force

2. Basic Usage

Enter the container

1
nerdctl exec -it hermes-agent bash

Set the model

1
/opt/hermes/.venv/bin/hermes model

Here I chose the custom model (○) 26. Custom endpoint (enter URL manually)

Chat

1
/opt/hermes/.venv/bin/hermes chat -q "hi"

Import OpenClaw

1
/opt/hermes/.venv/bin/hermes claw migrate

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
┌─────────────────────────────────────────────────────────┐
│          ⚕ Hermes — OpenClaw Migration                 │
└─────────────────────────────────────────────────────────┘


◆ Migration Settings
  Source:      /opt/data/.openclaw
  Target:      /opt/data
  Preset:      full
  Overwrite:   no (skip conflicts)
  Secrets:     yes (allowlisted only)

Diagnostics

1
/opt/hermes/.venv/bin/hermes doctor

Enter the TUI

1
/opt/hermes/.venv/bin/hermes

Add Telegram

1
2
echo "TELEGRAM_BOT_TOKEN=xxx
TELEGRAM_ALLOWED_USERS=xxx" >> /opt/data/.env

Add the user allowlist restriction.

1
/opt/hermes/.venv/bin/hermes gateway restart

3. Remote Invocation Integration

3.1 Calling with the CLI

Integration logic

Deploy a server on the deployment machine to expose the API; see https://github.com/shaowenchen/demo/tree/master/hermes-agent-call. After the server receives a request, it runs the following command on the host:

1
nerdctl exec -it hermes-agent /opt/hermes/.venv/bin/hermes chat --yolo -c -q "巡检一下集群"

The -c parameter here keeps a single session alive across calls.

Hosting the service with pm2

1
pm2 start "python3  hermes-agent-call.py" --name hermes-agent

How to call it

1
curl -X POST http://127.0.0.1:3011/call -H "Content-Type: application/json" -d '{"message":"hi"}'

1
{"ok": true, "sessionid": "", "message": "hi", "command": "nerdctl exec hermes-agent /opt/hermes/.venv/bin/hermes chat --yolo -c -q \"hi\"", "exitCode": 0, "stdout": "Hey chenshaowen! 👋\nWhat can I help you with today?", "stderr": ""}

3.2 Calling with the OpenAI API

Enable the OpenAI API

1
2
3
4
echo "API_SERVER_ENABLED=true
API_SERVER_HOST=0.0.0.0
API_SERVER_PORT=8643
API_SERVER_KEY=xxxx" >> /opt/data/.env

Restart Hermes Agent

1
nerdctl restart hermes-agent

How to call it

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
curl --location 'http://127.0.0.1:8643/v1/chat/completions' \
--header 'Content-Type: application/json' \
--header 'Authorization: Bearer xxx' \
--data '{
    "model": "default",
    "messages": [
        {
            "role": "user",
            "content": "你是谁？"
        }
    ],
    "temperature": 0.1,
    "max_tokens": 40960,
    "stream": false
}'

4. Viewing the Dashboard

The Hermes Agent dashboard cannot be used to chat directly for now; it is mainly for configuration and status checks.