FAQ

Frequently asked questions about Hermes Agent. If you can’t find an answer, feel free to ask on GitHub Issues.

Basic Questions

What are the minimum requirements for Hermes Agent?

Item	Minimum	Recommended
Python	3.10	3.11+
Memory	2GB	4GB+
Disk	1GB	5GB+
OS	Linux / macOS / WSL2	Ubuntu 22.04+
Network	LLM API access required	Broadband connection

Hermes Agent itself is very lightweight; most resource consumption comes from LLM API calls. If using local models (like Ollama), more memory is needed to run the models.

Which large language models does Hermes Agent support?

Hermes Agent supports 20+ LLM providers, including:

Type	Providers
International	OpenAI, Anthropic, Google Gemini, Mistral
Aggregators	OpenRouter (100+ models), Together AI, Fireworks AI
China Services	DeepSeek, Kimi (Moonshot), Qwen (Dashscope), Zhipu AI, SiliconFlow
Local Deployment	Ollama, vLLM, LM Studio
Cloud Services	Azure OpenAI, AWS Bedrock

All providers are accessed through a unified interface; use hermes model to interactively switch.

Is Hermes Agent free to use?

Hermes Agent itself is completely free (MIT open source license). However, you need to pay for LLM API calls:

• Lowest cost option: Use DeepSeek (approx ¥1/million tokens) or Kimi
• Free option: Use Ollama to run local models (requires decent hardware)
• Flexible option: Use OpenRouter (pay-as-you-go, supports free models)

Tip

If you just want to try it out, DeepSeek offers 5 million free tokens upon registration, more than enough to experience all features.

What is the relationship between Hermes Agent and OpenClaw?

Hermes Agent is an open source project developed by Nous Research; OpenClaw is a commercial product based on Hermes Agent. Think of it as:

• Hermes Agent = Open source community edition, fully featured, self-hosted
• OpenClaw = Commercial edition, provides hosted service, enterprise features, official support

See Hermes Agent vs OpenClaw for details.

Is my data secure?

Hermes Agent runs entirely on your device, data is stored locally in ~/.hermes/ directory:

• Conversation records stored locally, not uploaded to any third-party servers
• API Key stored in local configuration file
• Memory and skill data fully under your control

Caution

The only exception is LLM API calls — your messages are sent to the corresponding LLM provider for processing. For scenarios with strict data privacy requirements, we recommend using local models (Ollama / vLLM).

Feature Questions

Can I connect to multiple messaging platforms simultaneously?

Yes. Hermes Agent supports connecting to 15+ messaging platforms (Telegram, Discord, WeChat, Feishu, etc.) simultaneously, all platforms share the same Agent core memory and skills.

hermes chat --platform telegram --platform discord

Does it support voice interaction?

Yes. Use hermes chat --voice to enable voice mode, supporting voice input (STT) and voice output (TTS). You can configure the speech engine and language used.

How much can the memory system remember?

The memory system has no hard limit, but is affected by:

• Disk space — Memory files stored in ~/.hermes/memory/
• Token budget — In each conversation, historical memory occupies part of the context window
• Auto-summarization — Long-term memories are automatically compressed, retaining key information

Generally, several months of daily conversation history won’t cause any issues.

Can I customize the Agent’s personality?

Yes. Customize the Agent’s role, tone, and behavior guidelines through the SOUL.md file:

~/.hermes/SOUL.md

You are a humorous and witty technical consultant.
Answer questions concisely in English, occasionally throw in a cold joke.
Be rigorous with technical questions, don't make up uncertain information.

Can the Agent execute dangerous operations?

Hermes Agent has built-in safety mechanisms:

• Command approval — Dangerous operations (like deleting files) require user confirmation
• Sandbox execution — Code runs in sandbox
• User authorization — Can restrict which users can interact with the Agent
• Tool permissions — Can disable specific tools

See Security Best Practices for details.

Deployment Questions

Can it run on mobile phones?

Hermes Agent doesn’t run directly on mobile phones, but you can:

1. Deploy Hermes Agent on a VPS
2. Interact with the Agent through mobile apps like Telegram / Discord / WeChat

This is the most common usage pattern, the experience is almost identical to native mobile apps.

Can it run on Windows?

Hermes Agent doesn’t directly support Windows, but can run via:

• WSL2 (recommended) — Windows Subsystem for Linux
• Docker Desktop — Run using Docker container
• Virtual Machine — Run Linux virtual machine

What’s the difference between Docker deployment and production deployment?

No functional difference. Docker deployment advantages:

• Consistent environment, no system difference issues
• One-click startup, no manual dependency configuration
• Easy to scale and manage

Production environments recommend Docker Compose + systemd managed deployment, see VPS Deployment for details.

Can multiple people share one Hermes Agent?

Yes, through messaging platform group features:

• Add Bot to Telegram / Discord groups
• Configure user authorization to control who can use it
• Each user shares the same Agent memory and skills

Note

If multi-user independent memory is needed, deploy separate Hermes Agent instances for each user.

Performance Questions

What if response speed is slow?

1. Switch to faster model — Like Groq (ultra-fast inference) or DeepSeek
2. Reduce context length — Regularly /clear to empty conversation
3. Use local model — Ollama runs local models, no network latency
4. Check network — Confirm network latency to LLM API

What if token consumption is too fast?

Refer to Cost Optimization page for detailed strategies:

• Use cheaper models for simple tasks
• Enable auto-summarization to compress context
• Reasonably use fallback_models for routing

Other Questions

How to upgrade Hermes Agent?

# pip install
pip install --upgrade hermes-agent

# One-line script install
curl -fsSL https://raw.githubusercontent.com/NousResearch/hermes-agent/main/scripts/install.sh | bash

# Docker
docker pull nousresearch/hermes-agent:latest
docker compose up -d

# Source install
cd ~/.hermes-agent
git pull
pip install -e .

After upgrading, run hermes doctor to confirm everything is working.

How to backup and restore data?

# Backup
cp -r ~/.hermes ~/.hermes-backup-$(date +%Y%m%d)

# Restore
rm -rf ~/.hermes
cp -r ~/.hermes-backup-20260528 ~/.hermes

How to get help?

• 📖 Check this documentation site
• 🐛 GitHub Issues to report bugs
• 💬 Discord Community for discussions
• 📧 Attach hermes doctor diagnostic info when creating issues

Related Pages

• Quick Start — Get started in 5 minutes
• Common Errors — Common error solutions
• Cost Optimization — Reduce running costs
• Security Best Practices — Security hardening