How to update Hermes Agent and install skills from Telegram

Once Hermes Agent is running on your VPS with the Telegram gateway active, you have a full command interface in your pocket. You can update the agent, install community skills from agentskills.io, schedule recurring automations, and check the agent's health — all from Telegram, without touching a terminal.

• /help — Lists all available commands
• /new — Starts a fresh conversation context, clearing the current session
• /model — Shows your current AI model; request a switch in the same message
• /status — Shows session info: provider, model, memory state, and uptime
• /stop — Halts the currently running task
• /sethome — Designates this chat as the home channel for scheduled task results and proactive notifications

Before updating, ask your agent to back up its configuration files first: "Make a backup of my Hermes config files just in case." The agent will copy ~/.hermes/config.yaml and ~/.hermes/.env to safe timestamped backups. Once you have a backup, tell the agent: "Run hermes update to update yourself." Hermes will pull the latest changes from the official repository and reinstall all dependencies automatically. The gateway restarts when done and you get confirmation in the chat.

# 1. Ask the agent to back up config files first
cp ~/.hermes/config.yaml ~/.hermes/config.yaml.bak
cp ~/.hermes/.env ~/.hermes/.env.bak

# 2. Then update
hermes update

Hermes uses the open agentskills.io standard for skills. A skill is a folder containing a SKILL.md file with YAML frontmatter (name, description) and markdown instructions, plus optional scripts and resources. Once placed in ~/.hermes/skills/, the agent discovers the skill automatically the next time it starts. Hermes also auto-generates new skills during use: when it solves a problem in a reusable way, it writes that solution as a new skill file for next time — getting faster and more accurate on repeated tasks over time.

To install a community skill from Telegram, just describe what you need: "Find a skill for summarising long documents and install it." The agent will search agentskills.io and ClawHub, describe what it found, and install the skill into ~/.hermes/skills/ once you confirm. You can also hand it a URL directly: "Here is this skill from GitHub — read it and install it into my skills folder." This works well with Claude and GPT-4 models, which handle structured SKILL.md files especially cleanly. You can also create custom skills conversationally: "Create a skill so you remember how to format my weekly meeting notes." Hermes will write the skill file and use it automatically next time.

# Skills live here — auto-discovered on next session
~/.hermes/skills/

# Each skill follows the agentskills.io format:
~/.hermes/skills/my-skill/
  SKILL.md        # Required: name, description, instructions
  scripts/        # Optional: helper scripts
  resources/      # Optional: templates, assets

One of the most practical Telegram features in Hermes is natural-language cron scheduling. You do not write cron syntax — you just tell the agent what to do and when. Run /sethome in the Telegram chat first to designate where scheduled results are delivered, then describe your automation: "Every Monday at 9am, pull the top five AI news stories and send them here." Hermes schedules the task using its built-in cron system, which runs persistently on your VPS. You can set up multiple schedules — daily briefings, weekly reports, recurring backups — all without any cron configuration files.

• Run /sethome first to set the delivery channel for all scheduled output
• Describe schedules in plain English: "every weekday at 8am", "every Sunday night", "every hour"
• Layer multiple automations — daily news, weekly summaries, periodic health checks
• Scheduled tasks run on your VPS even when your phone is offline
• Ask the agent to list or cancel active schedules at any time

• /model in Telegram shows your current AI model and lets you request a switch
• Say "switch to claude-3-5-sonnet" — Hermes updates ~/.hermes/config.yaml and restarts the gateway
• Or use the ClawKickstart dashboard to reconfigure your model without downtime
• The Hermes gateway runs as a hermes-gateway systemd service — it survives reboots automatically

ClawKickstart provisions Hermes Agent with the Telegram gateway installed as a systemd service from day one — so it starts automatically after reboots and recovers from failures on its own. Your agent is one Telegram DM away from being updated, extended with new skills, scheduled to run recurring tasks, and re-targeted at whichever AI model you prefer.