NemoClaw User Guide - Setup Guide¶
Source: nvidia/skills (99 installs) Category: Security & Governance
A security-focused user guide for NemoClaw - a hardened OpenClaw variant providing permission models, audit logging, and secure execution boundaries for Hermes agents. Published by NVIDIA, signaling enterprise validation of the Hermes/OpenClaw agent ecosystem. Covers security best practices for deploying autonomous agents in production environments.
Installation¶
npx skills add nvidia/skills --skill nemoclaw-user-guide
Prerequisites¶
| Requirement | Details |
|---|---|
| Hermes Agent | v0.20.0+ with OpenClaw runtime support |
| NemoClaw Runtime | Installed alongside the skill - provides the hardened execution layer |
| Audit Log Storage | Local directory or S3 bucket for agent audit trails |
Roster Additions (Sep 27, 2026 sweep)¶
Two new nvidia/skills listings surfaced in the Sep 27 sweep:
| Skill | Installs | What It Does |
|---|---|---|
nemoclaw-user-get-started |
862 | NemoClaw onboarding and first-run workflow |
nemo-relay-install |
209 | NVIDIA Nemo relay component installation |
Install individually:
npx skills add nvidia/skills --skill nemoclaw-user-get-started
npx skills add nvidia/skills --skill nemo-relay-install
Roster Additions (Sep 29, 2026 sweep)¶
One additional nvidia/skills listing surfaced in the Sep 29 sweep:
| Skill | Installs | What It Does |
|---|---|---|
nemoclaw-user-overview |
828 | NemoClaw user overview and orientation |
Install individually:
npx skills add nvidia/skills --skill nemoclaw-user-overview
Key Capabilities¶
Security Features¶
| Capability | How to Trigger | Notes |
|---|---|---|
| Permission gates | Configured in ~/.hermes/config.yaml |
Fine-grained tool access control |
| Audit logging | Automatic - all agent actions logged | JSONL format, S3-compatible storage |
| Execution sandboxing | Enabled by default | Prevents unauthorized filesystem/network access |
| Security policy review | "Review my agent security policy" | Hardening recommendations |
| Threat detection | "Scan for security misconfigurations" | Checks permission model, env exposure, credential hygiene |
CLI Command Reference¶
# Run security audit on current agent config
hermes nemoclaw audit
# View recent security events
hermes nemoclaw events --last 24h
# Apply recommended hardening
hermes nemoclaw harden --apply
CorpusIQ Use Cases¶
| Use Case | How |
|---|---|
| Production Agent Security | Apply NemoClaw permission gates to CorpusIQ growth agents before production deployment |
| Compliance Auditing | Use audit logs for SOC 2 / security review evidence |
| Credential Hygiene | Automated scanning prevents credential leakage in agent output |
| Permission Hardening | Restrict agent tool access to minimum required for each workflow |
| Incident Response | Audit trails enable forensic analysis of agent actions |
Troubleshooting¶
| Issue | Fix |
|---|---|
| "Permission denied" on legitimate tool | Check ~/.hermes/config.yaml nemoclaw rules - add explicit allow |
| Audit log grows too large | Configure log rotation: nemoclaw config set --log-rotation 7d |
| "NemoClaw runtime not found" | Reinstall: npx skills add nvidia/skills --skill nemoclaw-user-guide --force |
Verification¶
# Verify skill installed
hermes skills list | grep nemoclaw-user-guide
# Run security audit
hermes nemoclaw audit
# Check audit logging is active
hermes nemoclaw status
← Skills Catalog | June 28 Discovery → Powered by CorpusIQ
Integration with CorpusIQ¶
CorpusIQ is the connector layer that lets AI assistants read the systems a business already runs on - accounting, payments, CRM, email, files and analytics - and answer with the source named. It is read-only by design: it retrieves data and never writes to, or acts in, a connected system, which is what makes it safe to point at live accounts.
This page is part of the CorpusIQ directory, the reference we maintain for the MCP server and agent-skill ecosystem. Run alongside CorpusIQ, a third-party server like this one can be combined with a company's own business data in a single answer.