hermesclaw

Hermes Agent (NousResearch) running inside NVIDIA OpenShell.

NVIDIA built OpenShell to hardware-enforce AI agent behavior — blocking network egress, filesystem writes, and dangerous syscalls at the kernel level. HermesClaw is a community implementation that puts Hermes Agent inside the same sandbox. The agent gets its full capability stack while the OS enforces hard limits. If a skill goes rogue, the kernel stops it.

Table of Contents

• Architecture
• Quick Start
  • Path 1 — Docker (no NVIDIA hardware)
  • Path 2 — OpenShell Sandbox
• What OpenShell Enforces
• Policy Presets
• Hermes Features
• Skills Library
• Use Cases
• HermesClaw vs NemoClaw
• CLI Reference
• Personalise Hermes
• Project Structure
• Diagnostics & Testing
• Contributing
• Related Projects

Architecture

OpenShell intercepts every call to inference.local inside the sandbox and routes it to the configured backend. Hermes never knows it's sandboxed.

Quick Start

Path 1 — Docker (no NVIDIA hardware required)

All Hermes features work. No kernel-level sandbox enforcement — Docker isolation only.

Step 1 — Clone and configure

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

Edit .env and set MODEL_FILE to your model filename. Download a model into models/:

# Example — Qwen3 4B (2.5 GB):
curl -L -o models/Qwen3-4B-Q4_K_M.gguf \
  https://huggingface.co/bartowski/Qwen3-4B-GGUF/resolve/main/Qwen3-4B-Q4_K_M.gguf

Step 2 — Build the image

./scripts/setup.sh

Step 3 — Start llama-server on your host (Hermes connects to it via host.docker.internal)

macOS:

brew install llama.cpp
llama-server -m models/your-model.gguf --port 8080 --ctx-size 32768 -ngl 99 --log-disable

Linux (build instructions):

llama-server -m models/your-model.gguf --port 8080 --ctx-size 32768 -ngl 99

Note: --ctx-size 32768 is required. Hermes's system prompt alone is ~11k tokens — lower values cause context overflow on every query.

Step 4 — Start Hermes

docker compose up -d
docker exec -it hermesclaw hermes chat -q "hello"

Path 2 — OpenShell Sandbox (full hardware enforcement)

Requires Linux + NVIDIA GPU + OpenShell installed.

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

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

# Build image + register OpenShell policy and profile
./scripts/setup.sh

# Start llama-server on the host
llama-server -m models/your-model.gguf --port 8080 --ctx-size 32768 -ngl 99

# Start Hermes inside the sandbox
./scripts/start.sh

Or use the hermesclaw CLI:

./scripts/hermesclaw onboard       # check all prerequisites
./scripts/hermesclaw start         # start with default (strict) policy
./scripts/hermesclaw start --gpu --policy gateway  # GPU + messaging enabled
./scripts/hermesclaw chat "hello"  # one-shot message

What OpenShell Enforces

All four layers are enforced out-of-process — even a fully compromised Hermes instance cannot override them.

Policy Presets

Switch security posture without restarting the sandbox:

./scripts/hermesclaw policy-set strict      # inference only (default)
./scripts/hermesclaw policy-set gateway     # + Telegram + Discord
./scripts/hermesclaw policy-set permissive  # + web search + GitHub skills

Hermes Features Inside the Sandbox

Skills Library

Pre-built skills that encode recurring workflows. Install with one command, invoke via chat:

./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
./skills/install.sh --all               # install everything

After installing, invoke from chat or any connected messaging platform:

docker exec -it hermesclaw hermes chat -q "run research-digest"
# or in Telegram: "run the anomaly-detection skill"

Full index: skills/README.md

Use Cases

Seven end-to-end guides covering real deployment scenarios — each with prerequisites, setup steps, automated tests, and a NemoClaw comparison:

Full index and NemoClaw compatibility table: docs/use-cases/

HermesClaw vs NemoClaw

Full comparison and test results: docs/test-results.md · docs/test-results-uc.md

hermesclaw CLI

hermesclaw onboard                    First-time setup and prerequisite check
hermesclaw start [--gpu] [--policy]   Start sandbox (OpenShell) or docker compose
hermesclaw stop                       Stop sandbox (memories + skills preserved)
hermesclaw status                     Show inference config + memory/skill counts
hermesclaw connect                    Open interactive shell inside sandbox
hermesclaw logs [--follow]            Stream sandbox logs
hermesclaw policy-list                List available policy presets
hermesclaw policy-set PRESET          Hot-swap policy without restart
hermesclaw doctor                     End-to-end diagnostic
hermesclaw chat "prompt"              One-shot message to Hermes
hermesclaw version                    Print version
hermesclaw uninstall                  Remove Docker image (data preserved)

Personalise Hermes

cp configs/persona.yaml.example configs/persona.yaml

Edit configs/persona.yaml — set your name, role, expertise, ticker watchlist, and response style. Hermes loads this into every session. For deeper personalisation, edit ~/.hermes/SOUL.md — this goes directly into the system prompt.

Project Structure

hermesclaw/
├── Dockerfile                          # Hermes Agent on debian:bookworm-slim
├── docker-compose.yml                  # Hermes container (llama-server runs on host)
├── .env.example                        # MODEL_FILE, CTX_SIZE, bot tokens
├── openshell/
│   ├── hermesclaw-policy.yaml          # Default policy
│   ├── hermesclaw-profile.yaml         # Sandbox profile
│   ├── policy-strict.yaml             # Inference only
│   ├── policy-gateway.yaml            # Inference + Telegram + Discord
│   └── policy-permissive.yaml         # Everything
├── configs/
│   ├── hermes.yaml.example            # Full Hermes config
│   └── persona.yaml.example           # User persona
├── skills/
│   ├── install.sh                     # Skill installer
│   ├── anomaly-detection/             # DB anomaly detection (detect.py)
│   ├── market-alerts/                 # Price threshold alerts (monitor.py)
│   ├── code-review/                   # Code review prompts
│   ├── slack-support/                 # FAQ + escalation bot
│   ├── home-assistant/                # HA MCP control
│   └── research-digest/               # Weekly arXiv digest
├── scripts/
│   ├── hermesclaw                     # Main CLI
│   ├── setup.sh                       # One-time setup
│   ├── start.sh / stop.sh / status.sh
│   ├── doctor.sh                      # End-to-end diagnostic
│   ├── test.sh                        # Feature comparison test suite
│   ├── test-setup.sh                  # Use-case test environment setup
│   └── test-uc-01.sh … test-uc-07.sh  # Per-use-case automated tests
├── docs/
│   ├── use-cases/                     # 7 end-to-end use-case guides
│   ├── features.md                    # Full feature reference
│   ├── test-results.md                # Feature comparison table
│   └── test-results-uc.md             # Use-case test results (2026-03-31)
├── knowledge/                         # Drop documents here (RAG context, read-only mount)
└── models/                            # Drop .gguf model weights here

Diagnostics & Testing

# Check your environment
./scripts/doctor.sh           # full diagnostic
./scripts/doctor.sh --quick   # skip slow checks

# Run the feature test suite
./scripts/test.sh             # generates docs/test-results.md
./scripts/test.sh --quick     # skip live inference tests

# Run use-case tests
bash scripts/test-setup.sh          # verify environment
bash scripts/test-uc-01.sh          # researcher
bash scripts/test-uc-04.sh          # data analyst (Postgres + anomaly detection)
bash scripts/test-uc-07.sh          # trader (latency measurement)

Contributing

HermesClaw welcomes contributions — especially:

• OpenShell policy corrections — if you have access to a real OpenShell environment, correctness fixes are the highest-value contribution
• New policy presets — homeassistant, coding, research, etc.
• New skills — follow the SKILL.md format in any existing skill as a template
• Real-world test reports — if you've run HermesClaw on NVIDIA hardware, share your ./scripts/doctor.sh output

Quick contributor setup:

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

Full guide: CONTRIBUTING.md · Code of Conduct · Changelog

Related

• hermes-agent-nemoclaw-openclaw — The parent repo: Hermes + NemoClaw + lightweight bots in one stack
• Hermes Agent — NousResearch's agent (18k ⭐)
• NemoClaw — NVIDIA's OpenClaw + OpenShell reference implementation
• OpenShell — NVIDIA's hardware-enforced AI sandbox