Module 10: Advanced Use Cases — AI Agents: Master Hermes Agent

**Course:** Hermes Agent — From Zero to Autonomous Agent

**Module:** 10 of 10

**Reading time:** ~25 minutes

**Difficulty:** Advanced

TL;DR

Use Case	What It Does	Key Tools
Trading Bot	Automated crypto/stock monitoring and execution	Cron jobs, API calls, terminal
Content Automation	AI script → voiceover → video → upload pipeline	Skills, terminal, API calls
Code Review Bot	Auto-review PRs via webhooks or cron	Webhooks, terminal, cron
Customer Support Agent	Automated support across Telegram/Discord	Gateway, allowlists, skills
Research Assistant	Daily digests, paper summaries, trend analysis	Cron, skills, web search
Personal Assistant	Email triage, calendar, todo management	API server, skills, cron
Multi-Agent Systems	Delegated task execution with sub-agents	Delegation, `/background`
Home Automation	Voice-controlled smart home via Home Assistant	Home Assistant adapter
CI/CD Integration	Automated deployment pipeline assistant	Webhooks, terminal
Data Pipeline	ETL, reporting, visualization automation	Cron, skills, Python execution

Use Case 1: Trading Bot

Build an automated trading bot that monitors markets, analyzes trends, and executes trades on your behalf.

Setup

You'll need API access to a trading platform. Hermes supports any exchange with a REST API via the terminal tool.

**Bybit example:**


# Add API keys to .env
BYBIT_API_KEY=your_api_key
BYBIT_API_SECRET=your_api_secret

Create a cron job for market monitoring:


/ Monitor BTC price every 30 minutes. If price drops >3% in 24h, 
  check RSI and volume. If RSI < 30 and volume > 2x average, 
  alert me on Telegram with a buy recommendation.

**Alpaca (stocks) example:**


# Add API keys to .env
ALPACA_API_KEY=your_api_key
ALPACA_SECRET_KEY=your_api_secret

Create a cron job:


Every morning at 9:30 AM EST, check my Alpaca portfolio. 
Show: current positions, P&L, top gainers/losers. 
If any position is down >5%, recommend whether to hold or sell.

Expected Results

Task	Output
Price monitoring	Alerts with technical analysis
Portfolio summary	Daily P&L report sent to Telegram
Backtesting	Historical performance analysis
Execution	Paper trading (always test before going live!)

Safety Considerations

- **Never give the agent full trading permissions** — Use read-only or whitelisted API keys

- **Paper trade first** — Run in simulation mode for at least a month

- **Set position limits** — Hard-code maximum trade sizes

- **Human-in-the-loop** — Use command approval for actual trades

- **Monitor costs** — Frequent API calls can add up

Use Case 2: Content Automation

Build an end-to-end content pipeline: Hermes writes a script, generates a voiceover, creates visuals, renders a video, and uploads it to YouTube.

The Pipeline


Cron job fires → Agent writes script → TTS generates voiceover
→ Python/moviepy renders video → Upload via YouTube API
→ Post link to Telegram/Discord

Setup

Create a content skill at `~/.hermes/skills/content-creator/SKILL.md`:


---
name: content-creator
description: Creates AI-generated video content from topic prompts
---

# Content Creator Guidelines

When creating a video from a topic:

1. Research the topic via web search
2. Write a 2-minute script (300 words, conversational tone)
3. Generate TTS audio using the configured voice
4. Find or generate relevant visuals
5. Use moviepy to compose the video with captions
6. Save to ~/content/output/

Cron Job

Create a daily content job:


Every morning at 6 AM, create a 2-minute explainer video about 
a trending AI news story from the past 24 hours. 
Use a professional, accessible tone. 
Save the video to ~/content/output/ and post the link to Telegram.

Required Tools

Tool	Purpose	Installation
ffmpeg	Audio/video processing	`sudo apt install ffmpeg`
moviepy	Python video editing	`pip install moviepy`
TTS	Voice generation	Built into Hermes via configured provider
OpenAI/Anthropic	Script writing	API key in `.env`

Example Workflow

1. Cron fires at 6 AM

2. Agent searches for top AI news

3. Writes a script with hook, body, and CTA

4. Generates voiceover via TTS

5. Creates captions/subtitles

6. Renders video with background music

7. Saves to output directory

8. Posts link to Telegram

Use Case 3: Code Review Bot

Automatically review pull requests in your GitHub repositories. Two approaches:

Approach A: Cron-Based (Simple)

Poll for open PRs on a schedule. See Module 7's GitHub PR review guide for the full setup.


# Cron job: every 2 hours, check for new PRs and review them
hermes cron create "0 */2 * * *" \
  "Check for new open PRs in myorg/backend-api.\n\n\
   For each PR opened or updated in the last 4 hours:\n\
   1. Fetch the diff\n\
   2. Review for bugs, security issues, code quality\n\
   3. Format findings with file:line, severity, and fix suggestion\n\n\
   If no new PRs, say: No new PRs to review." \
  --skill code-review \
  --deliver telegram

Approach B: Webhook-Based (Real-time)

Set up a webhook endpoint that GitHub pushes events to when PRs are opened or updated.

1. Configure the webhook adapter in `~/.hermes/gateway.json`:


{
  "platforms": {
    "webhook": {
      "enabled": true,
      "extra": {
        "port": 9000,
        "endpoints": {
          "/github-pr": {
            "description": "GitHub PR review webhook",
            "allowed_senders": ["github.com"],
            "script": "review_pr"
          }
        }
      }
    }
  }
}

2. Create a review script that Hermes executes when a webhook fires

3. Configure GitHub to send webhook events to `http://your-server:9000/github-pr`

**Pros:** No polling delay. Reviews happen seconds after a PR is opened.

**Cons:** Requires a public endpoint (port forwarding, or deploy on a cloud host).

Expected Results

- Every PR gets a review within minutes (webhook) or hours (cron)

- Reviews include specific line numbers and code snippets

- Team receives summary in Telegram/Discord

- Catch bugs before they reach production

Use Case 4: Customer Support Agent

Set up a Telegram or Discord bot that handles customer support for your product or service.

Setup

1. Create a skill with your product's knowledge base:


~/.hermes/skills/support/SKILL.md

Include:

- Product documentation and FAQs

- Known issues and workarounds

- Escalation criteria

- Brand voice and tone guidelines

2. Configure the bot with appropriate authorization:


# Allow all customers to message the bot
GATEWAY_ALLOW_ALL_USERS=true

# But restrict dangerous commands
# In config.yaml:
gateway:
  platforms:
    telegram:
      extra:
        allow_admin_from: ["your-user-id"]
        user_allowed_commands: [help, status]

Features

Feature	Implementation
FAQ answering	Skill with product knowledge
Ticket creation	Agent writes to your support system's API
Order lookup	API integration via terminal tool
Escalation	`/escalate` command starts a support ticket
Status checks	Cron job checks service health

Example


# User messages: "My order hasn't arrived in 2 weeks"
# Hermes checks the order system, finds the tracking info:
# "I can see order #12345 is with USPS (tracking: 1Z999AA10123456784).
#  Last scan was in Memphis distribution center. Would you like me to
#  contact USPS support? Or I can escalate this to our team."

Use Case 5: Research Assistant

An automated researcher that monitors topics, summarizes papers, and delivers digests.

Daily Digest


# Every morning at 7 AM
hermes cron create "0 7 * * *" \
  "Create a research digest:\n\n\
   1. Search arxiv for new papers about: large language models, AI agents\n\
   2. Search PubMed for: AI in healthcare\n\
   3. Search tech news for: AI industry developments\n\
   4. For each source, summarize the top 3 items\n\
   5. Format as a clean digest with sections and links\n\n\
   Focus on papers and news from the past 24 hours only." \
  --deliver telegram

Paper Summary on Demand


# Send this to your bot:
"Summarize this paper: [arxiv link].
 Focus on: methodology, results, limitations. 
 Compare it to the previous work in this area.
 End with: should I read the full paper? (yes/no + 1 sentence reason)"

Configurable Research Briefings

Create multiple cron jobs for different topics:


# AI research (weekdays only)
"Cron: 0 8 * * 1-5 — AI digest"

# Crypto (daily)
"Cron: 0 9 * * * — Crypto news and market analysis"

# Industry competitors (weekly)
"Cron: 0 10 * * 1 — Weekly competitor analysis"

Use Case 6: Personal Assistant

Use Hermes as a full personal assistant for email triage, calendar management, and task tracking.

Email Triage

Via the **API server** or **webhook** adapter, Hermes can receive and process emails:


1. Gateway receives email
2. Agent reads it and determines: important?, requires action?, spam?
3. Responds with: summary, priority, suggested action
4. Delivers digest to Telegram

Setup via email adapter:


# In .env
EMAIL_ENABLED=true
EMAIL_ADDRESS=assistant@yourdomain.com
EMAIL_PASSWORD=your_app_password
EMAIL_ALLOWED_SENDERS=trusted@example.com

Calendar & Todo

Hermes can manage a todo list in a skill file:


# Todo skill — stores tasks in ~/todo.md

When I say "remind me to X":
1. Add to ~/todo.md with date
2. Confirm with "Added: X"
3. Create a cron job if there's a time: "Remind me at [time]: X"

Daily check-in:


Every morning at 8 AM, read my todo list, check my calendar,
and give me a summary of what's scheduled and what's due today.

Use Case 7: Multi-Agent Systems

Hermes supports **delegation** — spawning sub-agents that work on tasks in parallel.

How Delegation Works

Within any session, you can tell Hermes to delegate tasks to sub-agents. Each sub-agent gets an isolated session and runs independently.


You: "Research three topics and combine the results"

Main agent:
  ├── → Sub-agent 1: "Research topic A"
  ├── → Sub-agent 2: "Research topic B"  
  └── → Sub-agent 3: "Research topic C"
  
  ← Collects all results
  ← Combines into one response

Use Cases

Task	Structure
Market research	Each competitor gets its own sub-agent
Code analysis	Each file/repo gets its own sub-agent
Writing	Outline → draft → review in parallel
Data processing	Each data source gets its own sub-agent

Via `/background`

The `/background` command spawns a separate agent instance:


/background Check all servers in the cluster and report any that are down

The background agent runs independently — you can keep chatting while it works. Results arrive in the same chat when complete.

Multi-Agent Kanban Board

Hermes ships with a **kanban plugin** for multi-agent orchestration:


hermes-plugin-kanban-install

This provides a visual board where you assign tasks to agents, track progress, and manage complex workflows.

Use Case 8: Home Automation via Home Assistant

Hermes integrates with **Home Assistant** to give you voice-controlled smart home management through Telegram or Discord.

Setup

Configure the Home Assistant platform:


# In .env
HOMEASSISTANT_ENABLED=true
HOMEASSISTANT_BASE_URL=http://192.168.1.100:8123
HOMEASSISTANT_ACCESS_TOKEN=your_long_lived_token

The Home Assistant adapter adds special tools:

- `ha_list_entities` — List all Home Assistant entities

- `ha_get_state` — Get the state of an entity

- `ha_call_service` — Call a Home Assistant service (turn on/off, set temp, etc.)

- `ha_list_services` — List available services

Examples


"Turn off all lights in the living room"
"Set the thermostat to 72 degrees"
"What's the temperature outside?"
"Lock the front door and arm the alarm"
"Good night" → Agent runs a scene: lights off, doors locked, alarm on

Automations with Cron


Every day at sunset: close the garage door if it's open
Every morning at 7 AM: turn on the coffee maker and set lights to 50%
When I leave home (detected by phone): turn off all lights and AC

Use Case 9: CI/CD Integration

Integrate Hermes into your CI/CD pipeline for automated deployments, monitoring, and rollback decisions.

Webhook-Triggered Deployments

When a CI/CD pipeline (GitHub Actions, GitLab CI, Jenkins) completes:

1. Pipeline sends webhook to Hermes

2. Hermes checks the build output, test results, and deployment logs

3. If tests pass and metrics are healthy → approve deployment

4. If tests fail or metrics degrade → notify the team with details

Monitoring Cron


Every 5 minutes, check the production server health:
- Is the app responding on port 443?
- Is the database connection pool under 80%?
- Is the error rate below 0.1%?
- Is memory usage under 80%?

If any check fails, alert the team in Telegram with the details.

Automated Rollback

When a deployment causes issues:


If error rate spikes above 5% within 10 minutes of a deployment:
1. Run: kubectl rollout undo deployment/app
2. Verify the previous version is serving traffic
3. Notify the team: "Rolled back deployment v1.2.3 due to error rate spike"

Use Case 10: Data Pipeline (ETL, Reporting, Visualization)

Use Hermes as an automated data processing engine that extracts, transforms, and visualizes data on a schedule.

ETL Pipeline


Cron job fires:
1. Extract: Download CSV from API or database
2. Transform: Clean data, compute metrics, aggregate
3. Load: Write results to a database or generate a report

Delivery:
- Send summary to Telegram
- Save full report to local files
- Generate charts and send them as images

Example: Sales Dashboard


hermes cron create "0 7 * * 1" \
  "Generate a weekly sales report:\n\n\
   1. Run: python3 ~/scripts/extract_sales_data.py\n\
   2. Calculate: total revenue, top products, YoY growth\n\
   3. Create a bar chart comparing this week vs last week\n\
   4. Create a pie chart of revenue by product category\n\
   5. Save charts to ~/weekly-reports/[date]/\n\
   6. Write a summary with key numbers and insights\n\
   7. Send the summary + charts to Telegram" \
  --deliver telegram

Visualization Tools

Tool	Purpose	Install
matplotlib	Python charts	`pip install matplotlib`
pandas	Data manipulation	`pip install pandas`
seaborn	Statistical charts	`pip install seaborn`
plotly	Interactive charts	`pip install plotly`
SQLite	Local database	Built into Python

Example: KPI Dashboard

A cron job that runs every hour:

1. Queries your database for key metrics

2. Checks them against thresholds

3. If any metric is outside acceptable range:

- Creates a chart showing the trend

- Sends an alert to Telegram with the chart

- Logs the incident for later review

Combining Use Cases

The real power of Hermes is combining multiple use cases into a single agent.

Example: Full-Tech-Stack Automation

A single Hermes instance running on a $6 VPS:

- **Trading bot** (monitors crypto markets, alerts on opportunities)

- **Code reviewer** (reviews team PRs every 2 hours)

- **CI/CD monitor** (watches deployments, handles rollbacks)

- **Home automation** (controls lights and thermostat via Telegram)

- **Daily briefing** (news, weather, server health every morning)

- **Research assistant** (weekly paper digest on AI topics)

- **Personal assistant** (todo list, reminders, email triage)

All from one gateway. All delivering to your Telegram.

The Limit Is Your Imagination

Hermes is a platform, not just a tool. The 10 use cases in this module are starting points — you can combine, extend, and customize them infinitely.

Some ideas to explore on your own:

- **Bot network** — Multiple Hermes instances, each specialized, communicating via webhooks

- **Customer onboarding** — Automated Telegram bot that walks new users through your product

- **Compliance monitoring** — Watch logs for PII leaks, security policy violations

- **Content moderation** — Auto-flag inappropriate content in your community

- **Language learning partner** — Chat with Hermes in your target language with corrections

- **Virtual D&D DM** — Generative dungeon master for your party

- **Personal tutor** — Analyzes your learning history and generates practice problems

- **Automated pentesting** — Scheduled security scans with report generation

Course Wrap-Up

You've reached the end of the Hermes Agent course. Let's recap what you've learned:

Module	Topic
:------:	-------
1	What Hermes is and why it matters
2	Installing Hermes on any platform
3	Configuration, providers, models
4	Basic usage, tools, terminal workflows
5	Building and training custom skills
6	Teaching and improving skills over time
7	Multi-platform messaging gateway
8	Production VPS deployment
9	Monetization strategies
10	Advanced use cases

Next Steps

1. **Deploy** — Set up Hermes on a VPS (Module 8)

2. **Connect** — Wire up Telegram or Discord (Module 7)

3. **Build** — Create your first custom skill (Module 5)

4. **Automate** — Schedule your first cron job (Module 6)

5. **Monetize** — Pick a strategy and start (Module 9)

6. **Explore** — Try the advanced use cases that interest you

Hermes is MIT-licensed, open-source, and maintained by Nous Research. The community is active on GitHub and Discord. Join us, share what you build, and help shape the future of autonomous AI agents.

Go build something amazing.

Advanced Use Cases

🎬 Full Walkthrough Video

TL;DR

Use Case 1: Trading Bot

Setup

Expected Results

Safety Considerations

Use Case 2: Content Automation

The Pipeline

Setup

Cron Job

Required Tools

Example Workflow

Use Case 3: Code Review Bot

Approach A: Cron-Based (Simple)

Approach B: Webhook-Based (Real-time)

Expected Results

Use Case 4: Customer Support Agent

Setup

Features

Example

Use Case 5: Research Assistant

Daily Digest

Paper Summary on Demand

Configurable Research Briefings

Use Case 6: Personal Assistant

Email Triage

Calendar & Todo

Use Case 7: Multi-Agent Systems

How Delegation Works

Use Cases

Via `/background`

Multi-Agent Kanban Board

Use Case 8: Home Automation via Home Assistant

Setup

Examples

Automations with Cron

Use Case 9: CI/CD Integration

Webhook-Triggered Deployments

Monitoring Cron

Automated Rollback

Use Case 10: Data Pipeline (ETL, Reporting, Visualization)

ETL Pipeline

Example: Sales Dashboard

Visualization Tools

Example: KPI Dashboard

Combining Use Cases

Example: Full-Tech-Stack Automation

The Limit Is Your Imagination

Course Wrap-Up

Next Steps