Hermes Agent Complete Backup Guide: How to Safely Back Up to GitHub (2026 Edition)

Hermes Agent (commonly known as Hermes Assistant / Hermes Lobster) is an open-source AI agent that self-evolves. Its core value lies in persistent memory and auto-generated skills. Once data is lost, months of accumulated memories and skills are gone forever.

This article will walk you through:

• Using built-in backup commands
• Securely uploading backups to GitHub
• Setting up automated backups (recommended)
• Complete restore process
• Advanced Git version control strategies

1. Where Does Hermes Store Its Data?

All important Hermes data is concentrated in this directory:

~/.hermes/

Key contents include:

config.yaml: Configuration

.env: API Keys (most sensitive)

skills/: All your skills (including auto-generated ones)

state.db: Memory and state database

sessions/: Session history

Other cache and authentication files

The essence of backup is to back up this directory.

2. Official Built‑in Backup Command (Most Recommended)

Hermes, from relatively recent versions, includes the hermes backup command, which works very well.

Full Backup (recommended for daily use)

hermes backup

After execution, a file like the following is generated:

~/hermes-backup-2026-07-01-041234.zip

Quick Backup (only core critical data, much faster)

hermes backup –quick

Suitable for multiple backups per day. Includes only core files like config.yaml, state.db, .env, auth, etc.

Advantages:

• Can be safely executed even while Hermes is running
• Uses SQLite native backup mechanism to ensure database consistency
• Automatically excludes runtime temporary files and checkpoints

3. Manual Backup to GitHub (Simplest Approach)

Steps:

1: Create a private repository on GitHub

Suggested repository name: hermes-backups or hermes-agent-backup

Make sure to select Private

2: Generate the backup

hermes backup

3: Upload to GitHub

Option 1: Drag and drop the zip file directly into the repository

Option 2: Create a Release and attach the zip as an asset (recommended for version management)

4: Add a description (optional but recommended)

Create a README.md in the repository root and write the backup date and description.

Pros: Simple and direct

Cons: Requires manual operation

4. Backup via Web Dashboard (Graphical Interface)

Run the following command to open the web‑based management interface:

hermes dashboard

On the Dashboard page, locate the Operations area. You will see:

Create backup → One‑click backup creation

Restore from backup → Restore from a backup

Some versions also support direct upload/download of backup files.

Suitable for users who prefer not to use the command line.

5. Automated Backup to GitHub (Recommended for Long‑term Use)

Option 1: Let Hermes schedule and push backups itself (smartest)

Simply tell Hermes: “Help me set up an automatic hermes backup --quick every day at 3 a.m., then push the generated zip file to my private GitHub repository hermes-backups.”

Hermes has a built‑in scheduled task system plus terminal execution capability, so it can handle the entire workflow (you need to configure Git and GitHub Token in advance).

Option 2: Use a community Skill (hermes-backup-repo)

There is a dedicated community Skill called hermes-backup-repo designed to:

• Automatically create and maintain a private backup repository on GitHub
• Periodically sync Hermes’s complete state (memory, skills, configuration, etc.)

After installing this Skill, you can achieve fairly comprehensive automated backups.

Option 3: Traditional Cron + Git (stable and reliable)

If you are familiar with Linux, write a simple script:

#!/bin/bash DATE=(date +%Y-%m-%d_%H%M%S) BACKUP_FILE=~/hermes-backup-DATE.zip

hermes backup –quick -o $BACKUP_FILE

Push to GitHub (Git must be configured beforehand)

cd ~/hermes-backups cp $BACKUP_FILE . git add . git commit -m “Hermes backup $DATE” git push origin main

Then use crontab -e to set up the scheduled task.

6. Complete Restore Process

Restoring on a new server or new computer is very simple:

1. Install Hermes Agent
2. Transfer the backup zip file to the new machine
3. Execute the restore command:

hermes import ~/hermes-backup-2026-07-01-041234.zip

4. After restore, it’s recommended to re‑run model configuration validation:

hermes model

Note: It is best to stop Hermes Gateway before restoring to avoid file conflicts.

7. Advanced Approach: Git Version Control for Key Files

If you want fine‑grained version control over your skills and memory, you can adopt this approach:

Place the following directories/files under Git management (using symlinks to point back):

• (~/.hermes/skills/)
• (~/.hermes/config.yaml)
• (Some important memory files)

This way, each skill update or important memory change can be committed via git commit, achieving real version history.

8. Important Considerations

Frequently Asked Questions (FAQ)

Q: What if the backup file is too large?

A: Use hermes backup --quick, or periodically clean up the cache/ directory and unneeded historical sessions.

Q: Can I back up to multiple places?

A: Yes. The most reliable approach is to back up simultaneously to GitHub, a local NAS, and a cloud drive.

Q: Will skills and memory survive a restore?

A: Yes. As long as you use the official hermes import, your skills and memory will be fully restored.

Q: How do I back up a Docker deployment?

A: Directly back up the mounted ~/.hermes directory, or use docker cp to export the entire container data volume.

Summary and Recommended Approach

Hermes’s value lies in the fact that “the more you use it, the better it understands you.” If data is lost, you reset an AI assistant that has accompanied you for months. Therefore, backups are really important. I recommend you run hermes backup today and upload the zip file to a private GitHub repository.