Plan a Safe Hermes Agent Setup

---
name: plan-safe-hermes-agent-setup
description: Create a safe, practical setup plan for a new Hermes Agent installation without collecting secrets in chat.
---

# Plan a Safe Hermes Agent Setup

Use this skill when a user wants to install or configure Hermes Agent for terminal, messaging, or agentic workflows.

## Inputs
- `operating_system`: Linux, macOS, WSL, or other environment
- `primary_interface`: CLI, Telegram, Discord, Slack, API, or mixed
- `model_provider`: OpenRouter, Anthropic, OpenAI, local model, or unknown
- `risk_level`: personal, team, production, or regulated
- `desired_capabilities`: tools, browser, terminal, MCP, cron, memory, gateway

## Procedure
1. Restate the user's target setup in one sentence.
2. Separate decisions into: model/provider, tools, memory, gateway, MCP, and automation.
3. Recommend the smallest working setup first.
4. Identify which steps require secrets and instruct the user to enter them locally, never in chat.
5. Add safety defaults: least privilege, explicit approvals for risky commands, public/private data separation, and secret redaction.
6. Define a verification checklist that proves the setup works.

## Output format
### Recommended starting setup
### Required decisions
### Safe configuration steps
### Verification checklist
### Common pitfalls

## Quality bar
Prefer an incremental, testable setup over a maximal configuration. Do not ask the user to paste API keys, passwords, private tokens, or proprietary files into the conversation.