A basic safety checklist for using Hermes at work
Hermes should be treated as an AI agent that can do real work, not as a simple chat tool. It can read and change files, run , automate browsers, call APIs, use MCP servers, remember context, schedule jobs, and act through connected accounts.
That power is why an IT or security team may block or limit it; this is normal information security, not just fear of AI. If Hermes can reach , credentials, company SaaS tools, client data, s, or outgoing messages, approval and controls are needed.
The useful official docs to check are security approvals and blocklists, tools and terminal backends, MCP behavior and trust, Managed Scope for organization-pinned settings, and Nous Portal/. This is practical workplace guidance, not legal, compliance, or policy advice.
Key points
- Hermes can edit files, run , automate browsers, call APIs, and schedule jobs.
- matters if Hermes can touch code, credentials, SaaS tools, client data, s, or external messages.
- Even a small report automation can become a problem if it runs through work accounts without approval.
- Start with the Hermes docs for approvals, blocklists, MCP trust, Managed Scope, and Nous Portal/.
- This is practical risk guidance, not legal or compliance advice.