### HermesClaw Onboarding and Setup

Source: https://context7.com/theaisingularity/hermesclaw/llms.txt

Guides through the initial setup by checking prerequisites and outlining next steps. The setup script is idempotent and safe to re-run.

```bash
hermesclaw onboard
```

```bash
./scripts/setup.sh
```

--------------------------------

### Install and Run Tailscale

Source: https://github.com/theaisingularity/hermesclaw/blob/main/docs/use-cases/03-home-automation/README.md

Install Tailscale on your home server to enable remote access. This command installs Tailscale and then starts it, assigning a stable IP address.

```bash
# Install Tailscale on your home server
curl -fsSL https://tailscale.com/install.sh | sh
sudo tailscale up
# Your server gets a stable IP like 100.x.x.x accessible from your phone
```

--------------------------------

### Start llama.cpp Server

Source: https://github.com/theaisingularity/hermesclaw/blob/main/docs/use-cases/06-privacy-regulated/README.md

Starts the llama.cpp inference server on the host, accessible via 'inference.local' from within the OpenShell sandbox. This setup is crucial for the strict network policy.

```bash
# CPU mode
docker run --rm -p 8080:8080 \
  -v ./models:/models:ro \
  ghcr.io/ggerganov/llama.cpp:server \
  -m /models/${MODEL_FILE} --port 8080 --host 0.0.0.0
```

--------------------------------

### Start HermesClaw and Install Slack Support Skill

Source: https://github.com/theaisingularity/hermesclaw/blob/main/docs/use-cases/05-small-business/README.md

Start the HermesClaw services in detached mode using Docker Compose and then install the slack-support skill.

```bash
docker compose up -d
./skills/install.sh slack-support
```

--------------------------------

### Install HermesClaw with One-Command Script

Source: https://context7.com/theaisingularity/hermesclaw/llms.txt

Installs HermesClaw by cloning the repository, pulling the Docker image, and setting up the CLI. Optionally override default paths using environment variables. After installation, download a model and start the llama-server.

```bash
# Install HermesClaw
curl -fsSL https://raw.githubusercontent.com/TheAiSingularity/hermesclaw/main/scripts/install.sh | bash
```

```bash
# Override defaults via environment variables
HERMESCLAW_HOME=~/ai/hermesclaw \
HERMESCLAW_BIN_DIR=~/.local/bin \
  curl -fsSL https://raw.githubusercontent.com/TheAiSingularity/hermesclaw/main/scripts/install.sh | bash
```

```bash
# After install: download a model (Qwen3 4B, ~2.5 GB)
curl -L -o ~/.hermesclaw/models/Qwen3-4B-Q4_K_M.gguf \
  https://huggingface.co/bartowski/Qwen3-4B-GGUF/resolve/main/Qwen3-4B-Q4_K_M.gguf
```

```bash
# Start llama-server on the host (macOS)
brew install llama.cpp
llama-server -m ~/.hermesclaw/models/Qwen3-4B-Q4_K_M.gguf \
  --port 8080 --ctx-size 32768 -ngl 99
```

```bash
# Start HermesClaw (Docker mode)
cd ~/.hermesclaw && docker compose up -d
```

```bash
# Verify
hermesclaw chat "hello"
# Output: Hello! How can I help you today?
```

--------------------------------

### Install Skills Library Packages

Source: https://github.com/theaisingularity/hermesclaw/blob/main/README.md

Install pre-built skills to automate recurring workflows. Skills can be installed individually or all at once. After installation, skills can be invoked via chat or connected messaging platforms.

```bash
./skills/install.sh research-digest
./skills/install.sh code-review
./skills/install.sh anomaly-detection
./skills/install.sh market-alerts
./skills/install.sh slack-support
./skills/install.sh home-assistant
./skills/install.sh --all
```

--------------------------------

### Install HermesClaw with One Command

Source: https://github.com/theaisingularity/hermesclaw/blob/main/README.md

Installs the prebuilt image, clones the repository, and sets up the CLI. Requires Docker, Git, and Curl. Ensure Docker is running.

```bash
curl -fsSL https://raw.githubusercontent.com/TheAiSingularity/hermesclaw/main/scripts/install.sh | bash
```

--------------------------------

### Clone HermesClaw Repository and Setup Environment

Source: https://github.com/theaisingularity/hermesclaw/blob/main/docs/use-cases/README.md

Clones the HermesClaw repository and sets up the initial environment for Docker-mode guides. Remember to place your GGUF model file in the 'models/' directory.

```bash
git clone https://github.com/TheAiSingularity/hermesclaw
cd hermesclaw
cp .env.example .env
# Drop your .gguf model into models/
```

--------------------------------

### Clone and Start HermesClaw

Source: https://github.com/theaisingularity/hermesclaw/blob/main/docs/use-cases/02-developer/README.md

Clone the hermesclaw repository, navigate into the directory, and copy the example environment file. Ensure a code-capable GGUF model is downloaded and configured in the .env file, along with an appropriate context size for code review.

```bash
git clone https://github.com/TheAiSingularity/hermesclaw
cd hermesclaw
cp .env.example .env
```

```bash
# Set in .env:
MODEL_FILE=Qwen3-14B-Q4_K_M.gguf
CTX_SIZE=16384   # larger context for code review
```

```bash
docker compose up -d
docker compose ps   # wait until both services are "healthy"
```

--------------------------------

### Install Market Alerts Skill

Source: https://github.com/theaisingularity/hermesclaw/blob/main/docs/use-cases/07-trader/README.md

Install the 'market-alerts' skill using the provided installation script. This skill is essential for setting up automated market monitoring and alerts.

```bash
./skills/install.sh market-alerts
```

--------------------------------

### Install Hermes Skills Interactively

Source: https://github.com/theaisingularity/hermesclaw/blob/main/CHANGELOG.md

The `skills/install.sh` script provides an interactive way to install pre-built Hermes skills by copying them to `~/.hermes/skills/`.

```bash
#!/bin/bash

# ... script logic to copy skills ...

echo "Skills installed to ~/.hermes/skills/"

```

--------------------------------

### Test Docker Build and Run

Source: https://github.com/theaisingularity/hermesclaw/blob/main/CONTRIBUTING.md

Build the Docker image for HermesClaw, verify the installation of Hermes within the container, and test the full Docker Compose setup in CPU mode.

```bash
# Build the hermesclaw container image
docker build -t hermesclaw:latest .

# Verify Hermes is installed inside
docker run --rm hermesclaw:latest hermes version

# Full compose test (CPU mode)
cp .env.example .env
docker compose up -d
docker exec hermesclaw hermes status
docker compose down
```

--------------------------------

### Download Model and Start Services

Source: https://github.com/theaisingularity/hermesclaw/blob/main/README.md

Steps to download model weights, start the llama-server, and launch HermesClaw. Adjust model path and server parameters as needed. Note the large context size requirement.

```bash
# 1. Download a GGUF model (example: Qwen3 4B, ~2.5 GB)
curl -L -o ~/.hermesclaw/models/Qwen3-4B-Q4_K_M.gguf \
  https://huggingface.co/bartowski/Qwen3-4B-GGUF/resolve/main/Qwen3-4B-Q4_K_M.gguf

# 2. Start llama-server on the host  (macOS shown; Linux: build llama.cpp from source)
brew install llama.cpp
llama-server -m ~/.hermesclaw/models/Qwen3-4B-Q4_K_M.gguf --port 8080 --ctx-size 32768 -ngl 99

# 3. Start HermesClaw
cd ~/.hermesclaw && docker compose up -d
hermesclaw chat "hello"
```

--------------------------------

### Start HermesClaw in OpenShell Sandbox

Source: https://github.com/theaisingularity/hermesclaw/blob/main/README.md

Commands to start HermesClaw within the NVIDIA OpenShell sandbox. Includes options for GPU acceleration and different policy modes like 'gateway'. Assumes OpenShell and HermesClaw are installed.

```bash
# Install OpenShell (requires NVIDIA account)
curl -fsSL https://www.nvidia.com/openshell.sh | bash

# Install HermesClaw via the one-liner above, then:
cd ~/.hermesclaw
llama-server -m models/your-model.gguf --port 8080 --ctx-size 32768 -ngl 99 &
hermesclaw start                     # default: strict policy
hermesclaw start --gpu --policy gateway  # GPU + messaging enabled
hermesclaw chat "hello"
```

--------------------------------

### Quick Contributor Setup

Source: https://github.com/theaisingularity/hermesclaw/blob/main/README.md

Commands to quickly set up the HermesClaw project for contribution. Includes cloning the repository, verifying the environment, and running feature tests.

```bash
git clone https://github.com/TheAiSingularity/hermesclaw
cd hermesclaw
./scripts/doctor.sh --quick    # verify your environment
./scripts/test.sh --quick      # run the feature test suite
shellcheck scripts/hermesclaw  # lint before submitting
```

--------------------------------

### Invoke Installed Skill via Messaging Platform

Source: https://github.com/theaisingularity/hermesclaw/blob/main/README.md

Trigger an installed skill by sending a message through a connected platform, such as Telegram. This example demonstrates invoking the 'anomaly-detection' skill.

```bash
# or in Telegram: "run the anomaly-detection skill"
```

--------------------------------

### Install OpenShell

Source: https://github.com/theaisingularity/hermesclaw/blob/main/docs/use-cases/06-privacy-regulated/README.md

Installs OpenShell on a Linux host. Requires an NVIDIA account and follows official instructions. For macOS on Apple Silicon, Colima must be set up first.

```bash
# Requires NVIDIA account — follow official instructions
curl -fsSL https://www.nvidia.com/openshell.sh | bash
openshell --version   # confirm install
```

```bash
brew install colima docker
colima start --memory 16 --cpu 4
# Then install OpenShell per the macOS docs
```

--------------------------------

### Test Setup Script for Use Cases

Source: https://github.com/theaisingularity/hermesclaw/blob/main/CHANGELOG.md

This script, `scripts/test-setup.sh`, is part of the sequential comparison test infrastructure. It prepares the environment for running use case tests.

```bash
#!/bin/bash

# ... setup logic ...

echo "Test setup complete."

```

--------------------------------

### Install and Manage Skills

Source: https://context7.com/theaisingularity/hermesclaw/llms.txt

Install individual or all skills using the provided script. After installation, restart the Docker container to apply changes. Skills can be invoked via CLI or Docker exec.

```bash
# Install individual skills
./skills/install.sh research-digest     # weekly arXiv digest → Telegram
./skills/install.sh code-review         # local code review (CLI or VS Code ACP)
./skills/install.sh anomaly-detection   # daily DB anomaly detection → Slack/Telegram
./skills/install.sh market-alerts       # watchlist price alerts → Telegram
./skills/install.sh slack-support       # Slack support bot with knowledge base
./skills/install.sh home-assistant      # natural language smart home control

# Install all at once
./skills/install.sh --all

# List available skills
./skills/install.sh --list

# After install, restart Docker container to pick up new skills
docker compose restart hermesclaw

# Invoke via CLI
hermesclaw chat "run research-digest"
hermesclaw chat "run anomaly-detection"

# Invoke via Docker exec
docker exec -it hermesclaw hermes chat -q "run market-alerts"

# Schedule a skill on a cron
hermesclaw chat "Every Monday at 8am, run research-digest and send to Telegram"
hermesclaw chat "Run anomaly-detection daily at 7am"

# Hermes auto-creates skills after complex tasks:
# After any session with 5+ tool calls, Hermes creates a skill in ~/.hermes/skills/

```

--------------------------------

### Clone and Verify Project

Source: https://github.com/theaisingularity/hermesclaw/blob/main/CONTRIBUTING.md

Clone the repository and run diagnostic scripts to verify the project setup. The quick tests should complete without failures, though warnings may appear if optional dependencies are missing.

```bash
git clone https://github.com/TheAiSingularity/hermesclaw
cd hermesclaw

# Run diagnostics on the repo itself (no model needed)
./scripts/doctor.sh --quick

# Run the feature comparison test suite
./scripts/test.sh --quick
```

--------------------------------

### Install Home Assistant Skill

Source: https://github.com/theaisingularity/hermesclaw/blob/main/docs/use-cases/03-home-automation/README.md

Install the home-assistant skill for hermesclaw using the provided installation script.

```bash
./skills/install.sh home-assistant
```

--------------------------------

### Start Remote Sandbox

Source: https://context7.com/theaisingularity/hermesclaw/llms.txt

Initiates a remote sandbox on a GPU server and selects the gateway. The sandbox is then created on the remote machine.

```bash
openshell gateway start --remote user@gpu-server
openshell gateway select my-remote-gateway
hermesclaw start    # creates sandbox on remote machine
```

--------------------------------

### Install Hermes Skills

Source: https://github.com/theaisingularity/hermesclaw/blob/main/skills/README.md

Use the install script to add skills to your Hermes environment. You can install individual skills, multiple skills, or all available skills.

```bash
./skills/install.sh research-digest
```

```bash
./skills/install.sh code-review anomaly-detection
```

```bash
./skills/install.sh --all
```

```bash
./skills/install.sh --list
```

--------------------------------

### Run Full OpenShell Test Suite

Source: https://github.com/theaisingularity/hermesclaw/blob/main/CONTRIBUTING.md

Execute the complete test suite for HermesClaw when OpenShell is installed and NVIDIA hardware is available. This involves setup, starting the service, running doctor, performing a chat interaction, setting the policy, and stopping the service.

```bash
./scripts/setup.sh
```

```bash
./scripts/start.sh
```

```bash
./scripts/hermesclaw doctor
```

```bash
./scripts/hermesclaw chat "hello, verify you can respond"
```

```bash
./scripts/hermesclaw policy-set gateway
```

```bash
./scripts/hermesclaw stop
```

--------------------------------

### Get OpenShell version

Source: https://github.com/theaisingularity/hermesclaw/blob/main/docs/compatibility.md

Prints the installed OpenShell version to the console. Required for compatibility reporting.

```bash
openshell --version
```

--------------------------------

### Docker Compose Quick Start

Source: https://github.com/theaisingularity/hermesclaw/blob/main/docs/features.md

Basic commands to set up and run Hermesclaw using Docker Compose. Ensure the .env file is configured.

```bash
cp .env.example .env         # fill in MODEL_FILE and optional tokens
docker compose up            # CPU inference
docker compose --profile gpu up  # NVIDIA GPU inference
```

--------------------------------

### Get Docker version

Source: https://github.com/theaisingularity/hermesclaw/blob/main/docs/compatibility.md

Prints the installed Docker version to the console. Required for compatibility reporting.

```bash
docker --version
```

--------------------------------

### Setup Script Enhancements for HermesClaw

Source: https://github.com/theaisingularity/hermesclaw/blob/main/CHANGELOG.md

The `scripts/setup.sh` script now registers an OpenShell inference provider and route on the first run, mirroring NemoClaw's behavior. It also bootstraps the `.env` file from `.env.example` if it doesn't exist.

```bash
#!/bin/bash
# ... existing script content ...

# Register OpenShell inference provider and route if OpenShell is present
if command -v openshell &> /dev/null;
  then
    openshell provider add local-llama --url http://localhost:8080
    openshell route add local-llama --provider local-llama --inference-model "*"
fi

# Bootstrap .env from .env.example if .env does not exist
if [ ! -f .env ]; then
  cp .env.example .env
  echo ".env created from .env.example"
fi

# ... rest of the script ...

```

--------------------------------

### Start Hermes ACP Server

Source: https://github.com/theaisingularity/hermesclaw/blob/main/docs/use-cases/02-developer/README.md

Initiate the Hermes ACP (Agent Communication Protocol) stdio server. This can be done either by executing the command inside the Docker container or directly on your local machine if Hermes is installed globally.

```bash
docker exec -it hermesclaw hermes acp
```

```bash
hermes acp
```

--------------------------------

### Install Research Digest Skill

Source: https://github.com/theaisingularity/hermesclaw/blob/main/docs/use-cases/01-researcher/README.md

Install the research-digest skill by copying its files into the Hermes skills directory.

```bash
./skills/install.sh research-digest
```

--------------------------------

### Run Use-Case Environment Check

Source: https://context7.com/theaisingularity/hermesclaw/llms.txt

Checks the environment setup for use-case specific tests.

```bash
bash scripts/test-setup.sh
```

--------------------------------

### Install Hermes Agent with a specific branch

Source: https://github.com/theaisingularity/hermesclaw/blob/main/docs/compatibility.md

Installs the Hermes Agent using a specified branch tag. This is the default installation method.

```bash
scripts/install.sh --branch <tag>
```

--------------------------------

### Hermes Agent Configuration Example

Source: https://context7.com/theaisingularity/hermesclaw/llms.txt

A comprehensive example of the Hermes agent configuration file, covering model selection, terminal settings, agent behavior, and personality definitions.

```yaml
# ~/.hermes/config.yaml

# Model / Inference
model:
  default: local                  # model ID from llama.cpp /v1/models
  provider: auto                  # auto | nous | openrouter | openai | anthropic | local
  base_url: http://inference.local  # overridden by OPENAI_BASE_URL env in Docker

  # Auxiliary models for specific tasks (optional)
  # vision_model: anthropic/claude-sonnet-4-6
  # memory_model: openai/gpt-4o-mini

# Terminal backend
terminal:
  backend: local
  timeout: 60

# Agent behavior
agent:
  max_turns: 90
  reasoning: medium               # low | medium | high
  approval_mode: smart            # manual | smart | off

  personalities:
    focused:
      description: "Fast, implementation-focused answers"
      system_prompt: "Be concise. Skip preamble. Give working code."
    researcher:
      description: "Deep research with citations"
      system_prompt: "Explore thoroughly. Cite sources. Show reasoning."
```

--------------------------------

### Configure Hermes.yaml for Slack Gateway

Source: https://github.com/theaisingularity/hermesclaw/blob/main/docs/use-cases/05-small-business/README.md

Copy the example Hermes configuration file and edit it to enable the Slack gateway and specify which tools are available via Slack.

```bash
cp configs/hermes.yaml.example configs/hermes.yaml
```

```yaml
gateway:
  slack:
    enabled: true
    # Restrict to specific channels or allow all DMs
    # allowed_channels: ["C0XXXXXX"]

tools:
  slack:
    enabled: [file, memory, skills]   # limit tools available via Slack gateway
```

--------------------------------

### Start Docker Stack (CPU and GPU Modes)

Source: https://github.com/theaisingularity/hermesclaw/blob/main/docs/use-cases/07-trader/README.md

Start the HermesClaw stack using Docker Compose. Use the default command for CPU mode or the `--profile gpu` flag if NVIDIA GPU acceleration is available.

```bash
docker compose up -d
```

```bash
docker compose --profile gpu up -d
```

--------------------------------

### Personalize Hermes Configuration

Source: https://github.com/theaisingularity/hermesclaw/blob/main/README.md

Copy the example persona configuration file to start customizing Hermes. Edit the copied file to set your name, role, expertise, and response style. For deeper personalization, modify the SOUL.md file which directly impacts the system prompt.

```bash
cp configs/persona.yaml.example configs/persona.yaml
```

--------------------------------

### Start HermesClaw Sandbox

Source: https://context7.com/theaisingularity/hermesclaw/llms.txt

Starts the HermesClaw sandbox using OpenShell or Docker Compose. Supports GPU acceleration and security policy presets. The sandbox name can be overridden using the HERMESCLAW_SANDBOX environment variable.

```bash
# Default: strict policy, no GPU
hermesclaw start
```

```bash
# GPU-accelerated inference with gateway policy (Telegram + Discord enabled)
hermesclaw start --gpu --policy gateway
```

```bash
# Permissive policy (web search + GitHub + all messaging)
hermesclaw start --policy permissive
```

```bash
# Environment variable: override the sandbox name (default: hermesclaw-1)
HERMESCLAW_SANDBOX=hermesclaw-dev hermesclaw start
```

```bash
# Expected output:
# Starting HermesClaw (OpenShell mode)...
#   Policy:  strict (/path/to/openshell/policy-strict.yaml)
#   Sandbox: hermesclaw-1
# ✓ Sandbox started: hermesclaw-1
```

--------------------------------

### Install Code-Review Skill

Source: https://github.com/theaisingularity/hermesclaw/blob/main/docs/use-cases/02-developer/README.md

Install the 'code-review' skill for HermesClaw by running the provided installation script. This action places the skill into the specified local directory, which is then mounted into the Docker container.

```bash
./skills/install.sh code-review
```

--------------------------------

### Install Anomaly Detection Skill

Source: https://github.com/theaisingularity/hermesclaw/blob/main/docs/use-cases/04-data-analyst/README.md

Install the anomaly-detection skill and its helper script into the local Hermes skills directory.

```bash
./skills/install.sh anomaly-detection
```

--------------------------------

### Register Profile and Start Sandbox

Source: https://github.com/theaisingularity/hermesclaw/blob/main/docs/use-cases/06-privacy-regulated/README.md

Registers the hermesclaw profile with OpenShell and then starts the sandbox with the 'strict' policy, enforcing data containment and network restrictions.

```bash
./scripts/setup.sh
# Registers hermesclaw profile with OpenShell
```

```bash
hermesclaw start --policy strict
# Creates sandbox hermesclaw-1 with the strict policy
```

--------------------------------

### Start Hermes Services

Source: https://github.com/theaisingularity/hermesclaw/blob/main/docs/use-cases/01-researcher/README.md

Start the Docker services for Hermes and its components in detached mode.

```bash
docker compose up -d
```

```bash
docker compose ps   # both should show "healthy"
```

--------------------------------

### Clone and Configure HermesClaw

Source: https://github.com/theaisingularity/hermesclaw/blob/main/docs/use-cases/05-small-business/README.md

Clone the repository, navigate to the directory, and copy the example environment file. Set the model file and Slack bot token in the .env file.

```bash
git clone https://github.com/TheAiSingularity/hermesclaw
cd hermesclaw
cp .env.example .env
```

```bash
MODEL_FILE=Qwen3-7B-Q4_K_M.gguf
SLACK_BOT_TOKEN=        # filled in step 2
```

--------------------------------

### Use Case Test Script Example

Source: https://github.com/theaisingularity/hermesclaw/blob/main/CHANGELOG.md

Example of a test script for a specific use case (e.g., `test-uc-01.sh`). These scripts are part of the sequential comparison test infrastructure and write results to `docs/test-results-uc.md`.

```bash
#!/bin/bash

# ... test execution logic ...

echo "Test for UC-01 completed."

```

--------------------------------

### Start Docker Compose and Verify Connection

Source: https://github.com/theaisingularity/hermesclaw/blob/main/docs/use-cases/04-data-analyst/README.md

Start the Docker services in detached mode and verify the database connection by listing tables using the Hermes chat command.

```bash
docker compose up -d
docker compose ps   # wait for "healthy"
```

```bash
docker exec -it hermesclaw hermes chat -q "List the tables in the database"
```

--------------------------------

### Install HermesClaw Skills

Source: https://github.com/theaisingularity/hermesclaw/blob/main/docs/use-cases/README.md

Installs specific Hermes skills for different use cases from the repository root. Ensure you are in the repository root directory before running these commands.

```bash
./skills/install.sh research-digest
./skills/install.sh code-review
./skills/install.sh home-assistant
./skills/install.sh anomaly-detection
./skills/install.sh slack-support
./skills/install.sh market-alerts
```

--------------------------------

### Get host OS and kernel information

Source: https://github.com/theaisingularity/hermesclaw/blob/main/docs/compatibility.md

Prints detailed host operating system and kernel information. Essential for compatibility reporting.

```bash
uname -a
```

--------------------------------

### Configure Inference Providers

Source: https://github.com/theaisingularity/hermesclaw/blob/main/docs/features.md

Commands to create and configure inference providers for various services like local llama.cpp, NVIDIA, OpenAI, Anthropic, and Ollama. Use `--from-existing` for cloud providers and specify credentials and base URLs for local setups.

```bash
openshell provider create --name local-llama --type openai \
  --credential OPENAI_API_KEY=not-needed \
  --config OPENAI_BASE_URL=http://127.0.0.1:8080/v1
openshell provider create --name nvidia-prod --type nvidia --from-existing
openshell provider create --name openai-prod --type openai --from-existing
openshell provider create --name anthropic-prod --type anthropic --from-existing
openshell provider create --name local-ollama --type openai \
  --credential OPENAI_API_KEY=dummy \
  --config OPENAI_BASE_URL=http://host.openshell.internal:11434/v1
```

--------------------------------

### Build HermesClaw from Source

Source: https://context7.com/theaisingularity/hermesclaw/llms.txt

Clones the repository, sets up the environment, and builds the Docker image. For OpenShell integration on Linux with NVIDIA GPUs, run the NVIDIA OpenShell installer script first.

```bash
git clone https://github.com/TheAiSingularity/hermesclaw
cd hermesclaw
cp .env.example .env           # Edit: set MODEL_FILE=Qwen3-4B-Q4_K_M.gguf
./scripts/setup.sh             # Builds hermesclaw:latest, creates ~/.hermes/config.yaml
```

```bash
# With OpenShell (Linux + NVIDIA GPU)
curl -fsSL https://www.nvidia.com/openshell.sh | bash   # requires NVIDIA account
hermesclaw start --policy strict
```

--------------------------------

### Start Docker Compose Services

Source: https://github.com/theaisingularity/hermesclaw/blob/main/docs/use-cases/03-home-automation/README.md

Launch all necessary services defined in the docker-compose.yml file in detached mode and verify their status.

```bash
docker compose up -d
docker compose ps   # wait for "healthy"
```

--------------------------------

### Enable Telegram Gateway in Hermes Config

Source: https://github.com/theaisingularity/hermesclaw/blob/main/docs/use-cases/01-researcher/README.md

Copy the example Hermes configuration and enable the Telegram gateway, specifying allowed users.

```bash
cp configs/hermes.yaml.example configs/hermes.yaml
```

```yaml
gateway:
  telegram:
    enabled: true
    # Restrict to your Telegram user ID (recommended)
    allowed_users: [YOUR_TELEGRAM_USER_ID]
```

```yaml
environment:
  HERMES_CONFIG: /sandbox/configs/hermes.yaml
```

--------------------------------

### Remote Deployment with HermesClaw

Source: https://github.com/theaisingularity/hermesclaw/blob/main/docs/features.md

Initiate a remote gateway and then start the HermesClaw agent on the remote machine. This allows for agent execution on remote infrastructure.

```bash
openshell gateway start --remote user@gpu-server
```

```bash
openshell gateway select my-remote-gateway
```

```bash
hermesclaw start   # now creates sandbox on remote machine
```

--------------------------------

### Copy Persona Configuration

Source: https://context7.com/theaisingularity/hermesclaw/llms.txt

Copy the example persona configuration to the active configuration file. Edit this file to personalize the agent's behavior and restart the Docker container for changes to take effect.

```bash
# Copy to configs/persona.yaml (gitignored)
# cp configs/persona.yaml.example configs/persona.yaml

name: "Alice Chen"
role: "ML Engineer"
expertise:
  - "Large Language Models"
  - "Systems engineering"
  - "Python and Rust"
interests:
  - "Local AI and privacy tech"
  - "Open-source software"
watchlist:          # Tickers for market summaries (used by market-alerts skill)
  - "NVDA"
  - "TSLA"
  - "AAPL"
style: "concise, technical, first-principles"
context: >
  I'm building an AI agent platform. Based in San Francisco.
  Prefer implementation-focused, data-driven answers.
  Current project: migrating inference stack to local llama.cpp.

```

```bash
# Apply persona
cp configs/persona.yaml.example configs/persona.yaml
# Edit persona.yaml, then restart:
docker compose restart hermesclaw

# For deeper personalization, edit the soul file directly:
# (mounted into container via hermesclaw-memories volume)
docker exec -it hermesclaw bash -c "cat /opt/hermes-data/SOUL.md"

```

--------------------------------

### Configure Postgres MCP Server

Source: https://github.com/theaisingularity/hermesclaw/blob/main/docs/use-cases/04-data-analyst/README.md

Copy the example Hermes configuration and add the Postgres MCP server details. Ensure alert channels are enabled and the service environment is correctly set in docker-compose.yml.

```yaml
mcp:
  servers:
    postgres:
      command: npx
      args:
        - "-y"
        - "@modelcontextprotocol/server-postgres"
        - "${DATABASE_URL}"
      timeout: 60
```

```yaml
gateway:
  slack:
    enabled: true     # if using Slack
  telegram:
    enabled: true     # if using Telegram
    allowed_users: [YOUR_TELEGRAM_USER_ID]
```

```yaml
environment:
  HERMES_CONFIG: /sandbox/configs/hermes.yaml
  DATABASE_URL: "${DATABASE_URL}"
```

--------------------------------

### HermesClaw CLI Commands

Source: https://github.com/theaisingularity/hermesclaw/blob/main/README.md

List of available commands for the HermesClaw command-line interface. Use these for setup, starting/stopping the agent, checking status, and interacting with the system.

```bash
hermesclaw onboard
hermesclaw start [--gpu] [--policy]
hermesclaw stop
hermesclaw status
hermesclaw connect
hermesclaw logs [--follow]
hermesclaw policy-list
hermesclaw policy-set PRESET
hermesclaw doctor
hermesclaw chat "prompt"
hermesclaw version
hermesclaw uninstall
```

--------------------------------

### Sandbox Lifecycle Management

Source: https://github.com/theaisingularity/hermesclaw/blob/main/docs/features.md

Commands for managing the sandbox lifecycle, including starting with specific configurations like GPU support or permissive policies, connecting to the sandbox, viewing logs, and monitoring the sandbox via a TUI.

```bash
hermesclaw start
hermesclaw start --gpu
hermesclaw start --policy permissive
hermesclaw connect
hermesclaw logs
hermesclaw logs --follow
openshell term
```

--------------------------------

### Switch Inference Backend and Model

Source: https://github.com/theaisingularity/hermesclaw/blob/main/docs/features.md

Commands to dynamically switch the active inference provider and model without restarting the sandbox. Use `inference get` to verify the current configuration.

```bash
openshell inference set --provider local-llama --model qwen3-4b
openshell inference set --provider nvidia-prod --model nemotron-4-340b-instruct
openshell inference set --provider anthropic-prod --model claude-sonnet-4-6
openshell inference get
```

--------------------------------

### Invoke Installed Skill via Chat

Source: https://github.com/theaisingularity/hermesclaw/blob/main/README.md

Execute an installed skill by invoking the 'hermes chat' command with the desired skill name. This example shows how to run the 'research-digest' skill.

```bash
docker exec -it hermesclaw hermes chat -q "run research-digest"
```

--------------------------------

### Clone Repository and Initial Configuration

Source: https://github.com/theaisingularity/hermesclaw/blob/main/docs/use-cases/07-trader/README.md

Clone the hermesclaw repository and set up the initial environment variables and model files. Ensure the correct model file is specified in the .env file and that the context size is adequate for the system prompt.

```bash
git clone https://github.com/TheAiSingularity/hermesclaw
cd hermesclaw
cp .env.example .env
```

```bash
# Download from HuggingFace (example URL pattern — use your preferred source)
# Filename: Qwen3-7B-Q4_K_M.gguf (~4 GB)
ls models/
```

```bash
MODEL_FILE=Qwen3-7B-Q4_K_M.gguf
CTX_SIZE=32768              # must be >= 32768 (Hermes system prompt is ~11k tokens)
TELEGRAM_BOT_TOKEN=         # filled in step 2
```

--------------------------------

### Build HermesClaw from Source

Source: https://github.com/theaisingularity/hermesclaw/blob/main/README.md

Instructions for building HermesClaw locally from source code. This involves cloning the repository, setting up environment variables, and building the Docker image.

```bash
git clone https://github.com/TheAiSingularity/hermesclaw
cd hermesclaw
cp .env.example .env                 # edit MODEL_FILE and any messaging tokens
./scripts/setup.sh                   # builds hermesclaw:latest locally
# ... then steps 2 and 3 above
```

--------------------------------

### Example Normal Daily Metrics Report

Source: https://github.com/theaisingularity/hermesclaw/blob/main/skills/anomaly-detection/SKILL.md

This is an example of the daily report when all metrics are within their normal range, indicating no anomalies detected.

```text
Daily Metrics — 2026-03-31 07:00

All clear — all metrics within normal range.

DAU:             8,432   (7d avg: 8,168  ↑ 3.2%)   ✅
Revenue:        $24,150  (7d avg: $24,420 ↓ 1.1%)   ✅
New signups:       312   (7d avg:    309  ↑ 1.0%)   ✅
Activation rate:  67.3%  (7d avg:  68.1% ↓ 0.8%)   ✅
```

--------------------------------

### Example Anomaly Alert Report

Source: https://github.com/theaisingularity/hermesclaw/blob/main/skills/anomaly-detection/SKILL.md

This is an example of an anomaly alert report, highlighting a specific metric that has deviated significantly from its baseline and suggesting possible causes.

```text
ANOMALY ALERT — 2026-03-31 07:00

Revenue: $15,200 (7d avg: $24,420 — z-score: −2.8) ⚠️
This is 38% below average. Possible causes:
- Payment processor issue
- Regional outage
- Recent pricing change
Check: SELECT * FROM payment_failures WHERE created_at >= CURRENT_DATE
```

--------------------------------

### Clone and Configure Hermes

Source: https://github.com/theaisingularity/hermesclaw/blob/main/docs/use-cases/01-researcher/README.md

Clone the Hermes repository and set up the initial environment configuration.

```bash
git clone https://github.com/TheAiSingularity/hermesclaw
cd hermesclaw
cp .env.example .env
```

```bash
# Example — Qwen3 4B (fast, good for research summarization)
# Download from HuggingFace or your preferred GGUF source
ls models/   # confirm .gguf file is present
```

```bash
MODEL_FILE=Qwen3-4B-Q4_K_M.gguf   # filename of your model in models/
```

--------------------------------

### Python SDK Initialization and Chat

Source: https://github.com/theaisingularity/hermesclaw/blob/main/docs/features.md

Initialize the AIAgent with specified models and toolsets, then perform single-turn chat interactions.

```python
from run_agent import AIAgent

agent = AIAgent(
    model="local",                         # or "anthropic/claude-opus-4-6"
    enabled_toolsets=["web", "terminal", "memory", "skills"],
    ephemeral_system_prompt="You are a helpful assistant.",
    max_iterations=90,
    skip_memory=False,                     # load MEMORY.md + USER.md
)

# Single-turn
response = agent.chat("Summarise the files in /sandbox")
print(response)
```

--------------------------------

### Get Full Sandbox State

Source: https://context7.com/theaisingularity/hermesclaw/llms.txt

Retrieves the detailed state of a specific sandbox.

```bash
openshell sandbox get hermesclaw-1
```

--------------------------------

### Run Quick Tests

Source: https://github.com/theaisingularity/hermesclaw/blob/main/CONTRIBUTING.md

Execute quick checks for the project before submitting a Pull Request. Ensure no new FAIL entries appear.

```bash
./scripts/doctor.sh --quick
./scripts/test.sh --quick
```

--------------------------------

### Test Pre-Market Briefing Generation

Source: https://github.com/theaisingularity/hermesclaw/blob/main/docs/use-cases/07-trader/README.md

Run a command to generate a pre-market summary for a watchlist using HermesClaw. This helps ensure the tool can provide timely market overviews.

```bash
docker exec -it hermesclaw hermes chat -q "Run the pre-market summary for my watchlist"
```

--------------------------------

### Get HermesClaw Version Information

Source: https://context7.com/theaisingularity/hermesclaw/llms.txt

Prints the version numbers for HermesClaw, OpenShell, and Hermes.

```bash
hermesclaw version
```

--------------------------------

### Restart HermesClaw in Docker

Source: https://github.com/theaisingularity/hermesclaw/blob/main/skills/README.md

After installing skills in Docker mode, restart the hermesclaw container to apply the changes.

```bash
docker compose restart hermesclaw
```

--------------------------------

### Validate YAML Files

Source: https://github.com/theaisingularity/hermesclaw/blob/main/CONTRIBUTING.md

This script validates all YAML files within the project to ensure they are syntactically correct. It requires the PyYAML library to be installed.

```python
import yaml, sys, glob
for f in glob.glob('**/*.yaml', recursive=True):
    try:
        yaml.safe_load(open(f))
        print(f'OK  {f}')
    except yaml.YAMLError as e:
        print(f'ERR {f}: {e}')
        sys.exit(1)
```

--------------------------------

### Setting up Hermes Messaging Gateway

Source: https://github.com/theaisingularity/hermesclaw/blob/main/docs/features.md

Configure the Hermes gateway within the sandbox for cross-platform messaging. This requires the 'gateway' network policy and can be set up interactively or via configuration files.

```bash
# Interactive setup (run inside sandbox)
hermesclaw connect
hermes gateway
```

```yaml
# Or configure manually in ~/.hermes/config.yaml
# Tokens in ~/.hermes/.env
```

--------------------------------

### Port Forwarding with HermesClaw

Source: https://github.com/theaisingularity/hermesclaw/blob/main/docs/features.md

Start port forwarding to allow communication between local and remote machines. Specify the local and remote ports to be forwarded.

```bash
openshell forward start hermesclaw-1 --local 9090 --remote 9090
```

--------------------------------

### Apply and Manage Policies

Source: https://github.com/theaisingularity/hermesclaw/blob/main/docs/features.md

Commands to apply, set, and view sandbox policies. Policies can be applied to running sandboxes for hot-reloading or globally. Use `--full` for detailed policy information and `list` to view history.

```bash
hermesclaw policy-set gateway
openshell policy set hermesclaw-1 --policy openshell/policy-gateway.yaml --wait
openshell policy set --global --policy openshell/policy-strict.yaml
openshell policy get hermesclaw-1 --full
openshell policy list hermesclaw-1
```

--------------------------------

### Clone and Configure Hermesclaw

Source: https://github.com/theaisingularity/hermesclaw/blob/main/docs/use-cases/03-home-automation/README.md

Clone the hermesclaw repository and set up the initial environment variables.

```bash
git clone https://github.com/TheAiSingularity/hermesclaw
cd hermesclaw
cp .env.example .env
```

--------------------------------

### Check Hermes version inside the container

Source: https://github.com/theaisingularity/hermesclaw/blob/main/docs/compatibility.md

Retrieves the Hermes version installed within the HermesClaw container. Useful for debugging and compatibility checks, available from v0.3.3.

```bash
cat /etc/hermes-version
```

--------------------------------

### Run Quick Feature Test Suite

Source: https://context7.com/theaisingularity/hermesclaw/llms.txt

Executes the feature comparison test suite, skipping live inference tests for faster execution. Useful for quick checks.

```bash
./scripts/test.sh --quick
```