Automated summary
Walks users through installing a NemoClaw agent on a Mac Mini with a dedicated macOS user account, an OpenShell container sandbox, and inference through Ollama Cloud.
Strengths
- Covers a full seven-part procedure, from host prerequisites (Colima, Docker, Ollama) to auto-start on boot via launchd
- Explains the dual-layer security model (non-admin macOS user isolation plus a Landlock/seccomp container) and the related hardening practices
- Gives concrete Colima CPU/memory/disk sizing per Mac Mini RAM tier, capped at 50% of host memory
- Ships a troubleshooting table for common failures such as the Docker daemon being down or Ollama being unreachable from the sandbox
Limitations
- macOS on Apple Silicon only; the dscl and launchd steps do not translate to other platforms
- Assumes NemoClaw and OpenShell are installable and that the free nemotron-3-super:cloud model is available through Ollama
- Depends on the bundled AGENT-SETUP.md and SECURITY.md files; SKILL.md alone does not include every command
Use it when someone wants to stand up one or more isolated NemoClaw agents on a Mac Mini and understand the security trade-offs of that setup.
Skip it for Linux deployments, remote clusters, or plain Ollama usage that does not need per-user OS isolation.
Security analysis
SafeThe skill provides guidance for setting up a sandboxed AI agent environment on macOS, including creating a dedicated user, installing Docker/Colima/Ollama, and configuring launchd. It does not instruct the AI to execute destructive, exfiltrating, or obfuscated commands. No tools are declared, so the skill only advises the user, who runs commands manually. While the setup involves system-level changes, the instructions are legitimate and focused on isolation and security.
No concerns found
How this analysis is madeExamples
Set up NemoClaw on my M2 Mac Mini with 16 GB RAM. I want a dedicated agent account called EVE and inference through Ollama Cloud.I already have one NemoClaw agent on my Mac Mini. Walk me through adding a second one with its own macOS user and sandbox, including how to pick the UNIQUE_ID.My NemoClaw sandbox says Ollama is not reachable and I also got a 'Cannot connect to Docker daemon' error. Help me diagnose both.NemoClaw Mac Mini Setup - Claude Skill
Description
Guide users through setting up a NemoClaw agent on a Mac Mini with a dedicated user account, dual-layer sandboxing (macOS user isolation + NemoClaw/OpenShell container), and Ollama Cloud inference.
When to Use This Skill
Invoke this skill when users ask about:
- Setting up NemoClaw on Mac Mini
- Creating isolated AI agent accounts on macOS
- Configuring NemoClaw with OpenShell
- NemoClaw security best practices
- Multiple agent setups on the same machine
- Ollama Cloud inference setup
- Sandboxed AI agent environments
Instructions
When a user requests NemoClaw setup assistance:
-
Understand their goal:
- Single agent or multiple agents?
- What's the agent name and username?
- What Mac Mini model do they have? (M1/M2/M4, RAM amount)
- Do they already have Colima/Docker installed?
- Do they already have Ollama installed?
-
Guide them through the setup:
- Reference the AGENT-SETUP.md guide
- Help them fill in the variables section:
- AGENT_NAME: Display name (e.g., "EVE", "WALL-E")
- AGENT_USERNAME: Lowercase username
- UNIQUE_ID: Auto-detect or specify (502, 503, etc.)
- ADMIN_USERNAME: Their admin username
-
Walk through each part:
- Part 1: Install host prerequisites (Colima, Docker, Ollama)
- Part 2: Create sandboxed macOS user (dscl commands)
- Part 3: Install NemoClaw as agent user (installer + onboarding wizard)
- Part 4: Verify setup (status, connect, test)
- Part 5: (Optional) Admin access and symlinks
- Part 6: (Optional) Auto-start on boot (launchd plists)
- Part 7: (Optional) Multiple agents
-
Help with onboarding wizard choices:
- Sandbox name: any name they like (default:
my-assistant) - Inference: Select option 2 (Local Ollama) — routes through local Ollama to cloud
- Model:
nemotron-3-super:cloud— free tier, runs via Ollama's cloud - Security policies: Accept defaults, customize later if needed
- Sandbox name: any name they like (default:
-
Security reminders:
- Explain the dual-layer security model (see SECURITY.md)
- Layer 1: macOS user isolation (no admin, no sudo)
- Layer 2: NemoClaw/OpenShell container (Landlock, seccomp, network namespace)
- Do NOT change ownership of NemoClaw config dirs
- Use
openshell termto monitor network egress - Recommend strong passwords for agent user accounts
- Recommend FileVault for disk encryption
-
Colima resource guidance:
- 16 GB Mac Mini:
--cpu 4 --memory 8 --disk 40 - 32 GB+ Mac Mini:
--cpu 6 --memory 12 --disk 60 - Don't allocate more than 50% of host RAM to Colima
- 16 GB Mac Mini:
-
Troubleshooting:
- "Cannot connect to Docker daemon" →
colima start - "nvidia-smi not found" → Expected on Apple Silicon, skip
- "Ollama not reachable from sandbox" → Check
brew services list | grep ollama - "User already exists" → Check/change UNIQUE_ID
- "Permission denied" → Re-run ownership commands
- Sandbox creation fails →
docker run hello-worldto test Docker
- "Cannot connect to Docker daemon" →
Example Usage
User: "How do I set up NemoClaw on my Mac Mini?"
Agent: "I'll help you set up NemoClaw on your Mac Mini with dual-layer sandboxing! This creates a dedicated user account for your agent and runs it inside NemoClaw's secure container.
First, let's gather some info:
- What would you like to name your agent? (e.g., 'EVE', 'WALL-E')
- Which Mac Mini model do you have? (M1/M2/M4, and how much RAM?)
- Do you already have Docker or Colima installed?
- Do you already have Ollama installed?
Once we have these, I'll walk you through the setup — it takes about 30-45 minutes and we'll set up:
- A sandboxed macOS user (Layer 1: OS isolation)
- NemoClaw with OpenShell (Layer 2: container sandboxing)
- Nemotron 3 Super via Ollama Cloud (free inference)
Let's get started!"
Files
AGENT-SETUP.md: Complete setup guide with all commandsSECURITY.md: Dual-layer security model explainedREADME.md: Quick overview, architecture diagram, and prerequisitesclaude-skill.md: This skill documentation
Repository
https://github.com/bcharleson/nemoclaw-macmini-setup
From the file to a first result
The download contains exactly the content displayed on this page. This copy may differ from the latest version in the source repository.
View the source and supporting filesThis download contains SKILL.md only. Any scripts, references, templates and dependencies are not included. Get the required files from the source, preserving their directory structure, before continuing.
1. Download the file
Download SKILL.mdFrom the root of a test project, save the file as .skill-downloads/nemoclaw-mac-mini-setup-guide-mryiqha7/SKILL.md, or use the command below. This folder is for inspecting the skill before installation.
Command for macOS/Linux or WSL: requires a POSIX shell and curl. On Windows without WSL, use the download button and create the folder manually.
mkdir -p '.skill-downloads/nemoclaw-mac-mini-setup-guide-mryiqha7' && curl --fail --show-error --location 'https://skills-guides.com/api/skills/nemoclaw-mac-mini-setup-guide-mryiqha7/download' --output '.skill-downloads/nemoclaw-mac-mini-setup-guide-mryiqha7/SKILL.md.part' && mv '.skill-downloads/nemoclaw-mac-mini-setup-guide-mryiqha7/SKILL.md.part' '.skill-downloads/nemoclaw-mac-mini-setup-guide-mryiqha7/SKILL.md'2. Check prerequisites
Read SKILL.md before using it. If it references scripts or external services, inspect them and prepare the required access in your test environment.
No recognized prerequisites section in this file. This does not mean it is self-contained: libraries, tools and access requirements may be mentioned elsewhere in the content or source repository.
The platforms on this page are catalog indications. The paths below follow tool documentation; they do not prove this skill has been run in those tools. No minimum tool version is confirmed for this listing.
3. Load the skill
This file does not provide a usable name/description header for this native procedure. It remains available to download and inspect; use the manual method below.
Manual method or another tool: attach SKILL.md to a conversation, or open it in an agent that can read files. Include any required supporting resources. No universal folder is recognized by every tool.
4. Try a first task
Use a short, synthetic input: an API endpoint, a small document or an interface brief. In a new conversation, attach the file or give the agent its path, then adapt this request:
Read .skill-downloads/nemoclaw-mac-mini-setup-guide-mryiqha7/SKILL.md and apply its instructions to this example: [describe a test input]. Identify any missing files, dependencies or access. Produce the result requested by the skill without inventing data or running unreviewed scripts.Expected result: output matching the skill instructions, or a precise list of missing prerequisites. A read confirmation alone does not validate a first use. If an external resource is unavailable, that path remains untested.
Configure a Chat Channel with SecretRefs
DevOps
Set up chat channels using non-interactive commands, with secrets only as SecretRefs. Covers gather, mutate, repair, and prove steps for reliable delivery.
Investigate ClickHouse CI Failures
DevOps
Investigate a ClickHouse CI failure end-to-end from a PR or S3 report URL. Fetches failed tests, classifies flaky vs real regression, and checks for existing issues/fixes.
Ansible PR Review
DevOps
Reviews an Ansible PR following the project's standardized process, including changelog checks, test coverage validation, and feedback tracking.